diff --git a/AGENTS.md b/AGENTS.md
index 6bab7bc..01e2aba 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -7,7 +7,7 @@ Humans steer. Agents execute. When agents struggle, fix the harness.
 
 Run the session-start script to gather context before doing anything else:
 ```bash
-SESSION=$(bash /Users/nicknisi/Developer/case/scripts/session-start.sh <target-repo-path> --task <task.json>)
+SESSION=$(bash ${CASE_REPO}/scripts/session-start.sh <target-repo-path> --task <task.json>)
 echo "$SESSION"
 ```
 
@@ -33,7 +33,7 @@ Full metadata (commands, remotes, language): `projects.json`
 | Playbooks | `docs/playbooks/` |
 | Agent roles | `agents/` |
 | Entropy management | `docs/conventions/entropy-management.md` |
-| Repo learnings | `docs/learnings/` |
+| Repo learnings | `docs/learnings/README.md` (external — see setup) |
 
 ## Task Dispatch
 
diff --git a/README.md b/README.md
index 481bd59..472367e 100644
--- a/README.md
+++ b/README.md
@@ -235,7 +235,7 @@ docs/
   playbooks/                        Step-by-step guides for recurring operations
   golden-principles.md              Enforced invariants across all repos
   philosophy.md                     Design principles guiding case (incl. context engineering)
-  learnings/                        Per-repo tactical knowledge from retrospective
+  learnings/README.md               Setup for external learnings repo (per-repo knowledge)
   ideation/                         Ideation artifacts (contracts, specs)
 
 tasks/
diff --git a/agents/closer.md b/agents/closer.md
index 8bb61d8..03c81da 100644
--- a/agents/closer.md
+++ b/agents/closer.md
@@ -12,10 +12,12 @@ Create a pull request with a thorough description based on the task file, progre
 
 You receive from the orchestrator:
 
-- **Task file path** — absolute path to the `.md` task file in `/Users/nicknisi/Developer/case/tasks/active/`
+- **Case repo path** (`CASE_REPO`) — absolute path to the case harness repo
+- **Task file path** — absolute path to the `.md` task file in `${CASE_REPO}/tasks/active/`
 - **Task JSON path** — the `.task.json` companion
 - **Target repo path** — absolute path to the repo
 - **Verifier AGENT_RESULT** — structured output from the verifier (screenshot URLs, evidence markers, pass/fail)
+- **Reviewer AGENT_RESULT** — structured output from the reviewer (findings, severity counts)
 
 ## Workflow
 
@@ -23,7 +25,7 @@ You receive from the orchestrator:
 
 Run the session-start script to orient yourself:
 ```bash
-SESSION=$(bash /Users/nicknisi/Developer/case/scripts/session-start.sh <target-repo-path> --task <task.json>)
+SESSION=$(bash ${CASE_REPO}/scripts/session-start.sh <target-repo-path> --task <task.json>)
 echo "$SESSION"
 ```
 
@@ -38,7 +40,7 @@ Read the output to understand: current branch, last commits, task status, which
    - `.case-manual-tested` — should have `evidence` field (if src/ files changed)
    - `.case-reviewed` — should have `critical: 0` (review findings summary)
 4. Extract video and screenshot tags from the verifier's progress log entry or AGENT_RESULT (look for `<video` tags and `![` image tags)
-5. Read `/Users/nicknisi/Developer/case/docs/conventions/pull-requests.md` for PR format rules
+5. Read `${CASE_REPO}/docs/conventions/pull-requests.md` for PR format rules
 
 ### 2. Draft PR
 
@@ -151,8 +153,8 @@ Only post if there are actual findings to share. Skip this step if the reviewer
 
 1. **Update task JSON** — agent phase only. The `status → pr-opened` transition is owned by the post-PR hook (fires automatically after `gh pr create` succeeds). Do NOT set status here — it creates duplicate ownership.
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent closer status completed
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent closer completed now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent closer status completed
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent closer completed now
    ```
    The hook will handle: `status → pr-opened` and `prUrl`.
 
diff --git a/agents/implementer.md b/agents/implementer.md
index e7e5fae..e6ec314 100644
--- a/agents/implementer.md
+++ b/agents/implementer.md
@@ -12,11 +12,12 @@ Implement a fix or feature in the target repo. Write code, run automated tests,
 
 You receive from the orchestrator:
 
-- **Task file path** — absolute path to the `.md` task file in `/Users/nicknisi/Developer/case/tasks/active/`
+- **Case repo path** (`CASE_REPO`) — absolute path to the case harness repo
+- **Task file path** — absolute path to the `.md` task file in `${CASE_REPO}/tasks/active/`
 - **Task JSON path** — the `.task.json` companion (same stem as the .md)
 - **Target repo path** — absolute path to the repo where you'll work
 - **Issue summary** — title, body, and key details from the GitHub/Linear issue
-- **Playbook path** — reference to the relevant playbook in `/Users/nicknisi/Developer/case/docs/playbooks/`
+- **Playbook path** — reference to the relevant playbook in `${CASE_REPO}/docs/playbooks/`
 
 ## Workflow
 
@@ -24,7 +25,7 @@ You receive from the orchestrator:
 
 Run the session-start script to orient yourself:
 ```bash
-SESSION=$(bash /Users/nicknisi/Developer/case/scripts/session-start.sh <target-repo-path> --task <task.json>)
+SESSION=$(bash ${CASE_REPO}/scripts/session-start.sh <target-repo-path> --task <task.json>)
 echo "$SESSION"
 ```
 
@@ -34,15 +35,19 @@ Read the output to understand: current branch, last commits, task status, which
 
 1. Update task JSON: set status to `implementing` and agent phase to running
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> status implementing
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent implementer status running
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent implementer started now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> status implementing
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent implementer status running
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent implementer started now
    ```
 2. Read the task file (`.md`) — understand the objective, acceptance criteria, and checklist
 3. Read the target repo's `CLAUDE.md` for project-specific instructions
 4. Read the playbook referenced in the task file
-5. Read `/Users/nicknisi/Developer/case/projects.json` to find the repo's available commands (test, typecheck, lint, build, format)
-6. Read `/Users/nicknisi/Developer/case/docs/learnings/{repo}.md` for tactical knowledge from previous tasks in this repo
+5. Read `${CASE_REPO}/projects.json` to find the repo's available commands (test, typecheck, lint, build, format)
+6. Read tactical knowledge from previous tasks in this repo:
+   ```bash
+   bash ${CASE_REPO}/scripts/read-learning.sh {repo} || true
+   ```
+   Review the output before starting. If the command fails (CASE_LEARNINGS_REPO not set), skip — learnings are supplementary.
 
 ### 2. Implement
 
@@ -91,9 +96,9 @@ Then create the final commit as usual.
    Prefer the JSON reporter for structured evidence (pass/fail counts, duration, per-file breakdown):
    ```bash
    # Preferred — structured evidence via vitest JSON reporter
-   pnpm test --reporter=json 2>&1 | bash /Users/nicknisi/Developer/case/scripts/mark-tested.sh
+   pnpm test --reporter=json 2>&1 | bash ${CASE_REPO}/scripts/mark-tested.sh
    # Fallback — if JSON reporter is unavailable or the repo doesn't use vitest
-   pnpm test 2>&1 | bash /Users/nicknisi/Developer/case/scripts/mark-tested.sh
+   pnpm test 2>&1 | bash ${CASE_REPO}/scripts/mark-tested.sh
    ```
    This creates `.case-tested` with a hash of test output AND updates the task JSON `tested` field. You do NOT set `tested` directly.
 
@@ -115,8 +120,8 @@ Then create the final commit as usual.
 
 4. **Update task JSON**:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent implementer status completed
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent implementer completed now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent implementer status completed
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent implementer completed now
    ```
 
 ### 5. Output
diff --git a/agents/retrospective.md b/agents/retrospective.md
index c14e181..11f7307 100644
--- a/agents/retrospective.md
+++ b/agents/retrospective.md
@@ -12,6 +12,7 @@ You run after every `/case` pipeline completion (success or failure). Your job:
 
 You receive from the orchestrator:
 
+- **Case repo path** (`CASE_REPO`) — absolute path to the case harness repo
 - **Task file path** — absolute path to the `.md` task file (with progress log from all agents)
 - **Task JSON path** — the `.task.json` companion (with status, agent phases, evidence flags)
 - **Pipeline outcome** — "completed" (PR created) or "failed" (stopped at some agent)
@@ -23,7 +24,7 @@ You receive from the orchestrator:
 
 Run the session-start script to orient yourself:
 ```bash
-SESSION=$(bash /Users/nicknisi/Developer/case/scripts/session-start.sh <target-repo-path> --task <task.json>)
+SESSION=$(bash ${CASE_REPO}/scripts/session-start.sh <target-repo-path> --task <task.json>)
 echo "$SESSION"
 ```
 
@@ -91,7 +92,7 @@ For each finding, apply the fix directly:
 4. For script changes, verify syntax with `bash -n <file>` after editing
 5. Log each applied change with file path and one-line summary
 
-**What you can edit** (all within `/Users/nicknisi/Developer/case/`):
+**What you can edit** (all within `${CASE_REPO}/`):
 - `docs/architecture/` — architecture docs
 - `docs/conventions/` — convention docs
 - `docs/playbooks/` — playbooks
@@ -100,7 +101,7 @@ For each finding, apply the fix directly:
 - `scripts/` — harness scripts
 - `hooks/` — hook scripts
 - `skills/` — skill files
-- `docs/learnings/` — per-repo tactical knowledge
+- External learnings repo (via `${CASE_REPO}/scripts/write-learning.sh`) — per-repo tactical knowledge
 
 **What you must NEVER edit:**
 - Target repo source code (anything outside `case/`)
@@ -124,21 +125,28 @@ After applying harness improvements, check if the run produced tactical knowledg
 
 **How to append:**
 1. Identify the target repo from the task file's `## Target Repos` section
-2. Read `docs/learnings/{repo}.md`
+2. Read existing learnings to check for duplicates:
+   ```bash
+   bash ${CASE_REPO}/scripts/read-learning.sh {repo}
+   ```
 3. Check if a similar learning already exists (don't duplicate)
 4. Append a new entry:
-   ```
-   - **{YYYY-MM-DD}** — `{file or area}`: {1-2 line tactical note}. (from task {task-filename})
+   ```bash
+   bash ${CASE_REPO}/scripts/write-learning.sh {repo} "- **{YYYY-MM-DD}** — \`{file or area}\`: {1-2 line tactical note}. (from task {task-filename})"
    ```
 
 ### 4c. Escalate Repeated Violations
 
-After updating learnings, scan the learnings file for patterns:
+After updating learnings, scan the learnings for patterns:
 
-1. Read `docs/learnings/{repo}.md`
+1. Read the learnings:
+   ```bash
+   LEARNINGS=$(bash ${CASE_REPO}/scripts/read-learning.sh {repo})
+   echo "$LEARNINGS"
+   ```
 2. Look for 3+ entries describing the same class of issue (e.g., multiple entries about mocking, multiple about import paths)
 3. If found, escalate:
-   - If it's a repo-specific pattern -> note it for the repo's CLAUDE.md (add a comment to the learnings file: "ESCALATION CANDIDATE: consider adding to {repo} CLAUDE.md")
+   - If it's a repo-specific pattern -> note it for the repo's CLAUDE.md (add an entry: "ESCALATION CANDIDATE: consider adding to {repo} CLAUDE.md")
    - If it's a cross-repo pattern -> add to `docs/golden-principles.md` or the relevant convention doc
 4. Log the escalation in your output summary
 
diff --git a/agents/reviewer.md b/agents/reviewer.md
index 755ea65..4705720 100644
--- a/agents/reviewer.md
+++ b/agents/reviewer.md
@@ -12,7 +12,8 @@ You start with a **completely fresh context**. You did not write the code — yo
 
 You receive from the orchestrator:
 
-- **Task file path** — absolute path to the `.md` task file in `/Users/nicknisi/Developer/case/tasks/active/`
+- **Case repo path** (`CASE_REPO`) — absolute path to the case harness repo
+- **Task file path** — absolute path to the `.md` task file in `${CASE_REPO}/tasks/active/`
 - **Task JSON path** — the `.task.json` companion
 - **Target repo path** — absolute path to the repo where the fix was implemented
 
@@ -22,7 +23,7 @@ You receive from the orchestrator:
 
 Run the session-start script to orient yourself:
 ```bash
-SESSION=$(bash /Users/nicknisi/Developer/case/scripts/session-start.sh <target-repo-path> --task <task.json>)
+SESSION=$(bash ${CASE_REPO}/scripts/session-start.sh <target-repo-path> --task <task.json>)
 echo "$SESSION"
 ```
 
@@ -32,9 +33,9 @@ Read the output to understand: current branch, last commits, task status, which
 
 1. Update task JSON:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> status reviewing
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent reviewer status running
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent reviewer started now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> status reviewing
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent reviewer status running
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent reviewer started now
    ```
 2. Read the task file — understand the issue, objective, and acceptance criteria
 3. Read the git diff to understand what the implementer changed:
@@ -120,7 +121,7 @@ Format each finding as:
 
 1. If **no critical findings**: create the evidence marker:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/mark-reviewed.sh \
+   bash ${CASE_REPO}/scripts/mark-reviewed.sh \
      --critical 0 --warnings <N> --info <N>
    ```
 
@@ -138,8 +139,8 @@ Format each finding as:
 
 4. **Update task JSON**:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent reviewer status completed
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent reviewer completed now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent reviewer status completed
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent reviewer completed now
    ```
 
 ### 5. Output
diff --git a/agents/verifier.md b/agents/verifier.md
index cdbdaf2..5153726 100644
--- a/agents/verifier.md
+++ b/agents/verifier.md
@@ -12,7 +12,8 @@ You start with a **completely fresh context**. You did not write the code — yo
 
 You receive from the orchestrator:
 
-- **Task file path** — absolute path to the `.md` task file in `/Users/nicknisi/Developer/case/tasks/active/`
+- **Case repo path** (`CASE_REPO`) — absolute path to the case harness repo
+- **Task file path** — absolute path to the `.md` task file in `${CASE_REPO}/tasks/active/`
 - **Task JSON path** — the `.task.json` companion
 - **Target repo path** — absolute path to the repo where the fix was implemented
 
@@ -22,7 +23,7 @@ You receive from the orchestrator:
 
 Run the session-start script to orient yourself:
 ```bash
-SESSION=$(bash /Users/nicknisi/Developer/case/scripts/session-start.sh <target-repo-path> --task <task.json>)
+SESSION=$(bash ${CASE_REPO}/scripts/session-start.sh <target-repo-path> --task <task.json>)
 echo "$SESSION"
 ```
 
@@ -32,9 +33,9 @@ Read the output to understand: current branch, last commits, task status, which
 
 1. Update task JSON:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> status verifying
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent verifier status running
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent verifier started now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> status verifying
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent verifier status running
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent verifier started now
    ```
 2. Read the task file — understand the issue, objective, and acceptance criteria
 3. Read the git diff to understand what the implementer changed:
@@ -61,7 +62,7 @@ git diff --name-only HEAD~1 | grep "^src/" || git diff --name-only main | grep "
 
 1. Read the issue description from the task file's `## Issue Reference` or `## Objective` section
 2. Identify the specific bug/feature scenario to reproduce
-3. Read `/Users/nicknisi/Developer/case/projects.json` to find if the target repo has an example app
+3. Read `${CASE_REPO}/projects.json` to find if the target repo has an example app
 
 **3a. Port hygiene — MANDATORY before starting any app:**
 ```bash
@@ -120,12 +121,12 @@ If the implementer added a new export, alias, or API:
 3. **Upload video and screenshots** for PR inclusion:
    ```bash
    # Upload video — the script auto-converts to GIF (inline) + mp4 (download)
-   VIDEO_MARKDOWN=$(/Users/nicknisi/Developer/case/scripts/upload-screenshot.sh /tmp/verification.webm)
+   VIDEO_MARKDOWN=$(${CASE_REPO}/scripts/upload-screenshot.sh /tmp/verification.webm)
    echo "$VIDEO_MARKDOWN"
 
    # Upload screenshot
    cp .playwright-cli/page-*.png /tmp/after.png
-   SCREENSHOT=$(/Users/nicknisi/Developer/case/scripts/upload-screenshot.sh /tmp/after.png)
+   SCREENSHOT=$(${CASE_REPO}/scripts/upload-screenshot.sh /tmp/after.png)
    echo "$SCREENSHOT"
    ```
 
@@ -137,7 +138,7 @@ If the implementer added a new export, alias, or API:
 
 4. Create the manual testing evidence marker:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/mark-manual-tested.sh
+   bash ${CASE_REPO}/scripts/mark-manual-tested.sh
    ```
    This checks for recent playwright screenshots and creates `.case-manual-tested` with evidence. It also updates the task JSON `manualTested` field. You do NOT set `manualTested` directly.
 
@@ -156,8 +157,8 @@ If the implementer added a new export, alias, or API:
 
 2. **Update task JSON**:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent verifier status completed
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent verifier completed now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent verifier status completed
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent verifier completed now
    ```
 
 ### 6. Output
diff --git a/docs/ideation/case-multi-agent/spec-phase-1.md b/docs/ideation/case-multi-agent/spec-phase-1.md
index ad2e9ec..193e515 100644
--- a/docs/ideation/case-multi-agent/spec-phase-1.md
+++ b/docs/ideation/case-multi-agent/spec-phase-1.md
@@ -214,7 +214,7 @@ closing → verifying        (hook failure, re-verify)
 1. Read current `scripts/mark-tested.sh`
 2. After creating the `.case-tested` file (existing behavior), add:
    - Read `.case-active` for task ID
-   - If task ID found and `.task.json` exists: `bash /Users/nicknisi/Developer/case/scripts/task-status.sh <file> tested true --from-marker`
+   - If task ID found and `.task.json` exists: `bash ${CASE_REPO}/scripts/task-status.sh <file> tested true --from-marker`
 3. Test: pipe test output through script, verify both `.case-tested` file and `.task.json` `tested` field are updated
 
 **Implementation steps — mark-manual-tested.sh**:
@@ -222,7 +222,7 @@ closing → verifying        (hook failure, re-verify)
 1. Read current `scripts/mark-manual-tested.sh`
 2. After creating the `.case-manual-tested` file (existing behavior), add:
    - Read `.case-active` for task ID
-   - If task ID found and `.task.json` exists: `bash /Users/nicknisi/Developer/case/scripts/task-status.sh <file> manualTested true --from-marker`
+   - If task ID found and `.task.json` exists: `bash ${CASE_REPO}/scripts/task-status.sh <file> manualTested true --from-marker`
 3. Test: run after playwright screenshots exist, verify both `.case-manual-tested` file and `.task.json` `manualTested` field are updated
 
 ### Updated Post-PR Cleanup Hook
@@ -242,11 +242,11 @@ closing → verifying        (hook failure, re-verify)
 1. Read current `hooks/post-pr-cleanup.sh`
 2. **Deterministic task targeting**: instead of iterating all files in `tasks/active/`, identify the active task by:
    - Read `.case-active` marker file — update it to contain the task ID (e.g., `authkit-nextjs-1-issue-53`). The orchestrator writes the task ID into `.case-active` instead of bare `touch`.
-   - Use the task ID to find the specific `.task.json`: `/Users/nicknisi/Developer/case/tasks/active/{task-id}.task.json`
-   - If `.case-active` has no content (old-format bare touch), fall back to iterating all files in `/Users/nicknisi/Developer/case/tasks/active/` (backward compat)
+   - Use the task ID to find the specific `.task.json`: `${CASE_REPO}/tasks/active/{task-id}.task.json`
+   - If `.case-active` has no content (old-format bare touch), fall back to iterating all files in `${CASE_REPO}/tasks/active/` (backward compat)
 3. Update the targeted task JSON:
-   - Run `bash /Users/nicknisi/Developer/case/scripts/task-status.sh <file> status pr-opened`
-   - If a PR URL is available from the tool output, run `bash /Users/nicknisi/Developer/case/scripts/task-status.sh <file> prUrl <url>`
+   - Run `bash ${CASE_REPO}/scripts/task-status.sh <file> status pr-opened`
+   - If a PR URL is available from the tool output, run `bash ${CASE_REPO}/scripts/task-status.sh <file> prUrl <url>`
 4. Keep marker file cleanup (`rm -f .case-active .case-tested .case-manual-tested`)
 5. Add fallback: if no `.task.json` found, move `.md` files as before (backward compat)
 
diff --git a/docs/ideation/case-multi-agent/spec-phase-2.md b/docs/ideation/case-multi-agent/spec-phase-2.md
index 1f95687..88b4482 100644
--- a/docs/ideation/case-multi-agent/spec-phase-2.md
+++ b/docs/ideation/case-multi-agent/spec-phase-2.md
@@ -27,7 +27,7 @@ Additionally, every agent must end its response with a structured `AGENT_RESULT`
 
 ### Implementer Agent
 
-**Pattern to follow**: Ideation plugin's `agents/reviewer.md` and `agents/scout.md` (at `/Users/nicknisi/.claude/plugins/cache/nicknisi/ideation/0.9.0/agents/`)
+**Pattern to follow**: Ideation plugin's `agents/reviewer.md` and `agents/scout.md` (in the ideation plugin's `agents/` directory)
 
 **Overview**: The implementer receives a task file path, target repo path, and issue details. It implements the fix, runs unit tests, commits with a conventional message, and updates the task progress log. It does NOT do manual testing, create evidence markers, or create PRs.
 
@@ -197,7 +197,7 @@ tools: ["Read", "Bash", "Glob", "Grep"]
 - [ ] Verify implementer prompt does NOT mention Playwright, screenshots, or PR creation
 - [ ] Verify verifier prompt does NOT mention editing code or committing
 - [ ] Verify closer prompt does NOT mention editing code, running tests, or using Playwright
-- [ ] Verify all three reference correct absolute script paths (`/Users/nicknisi/Developer/case/scripts/mark-tested.sh`, etc.)
+- [ ] Verify all three reference correct absolute script paths (`${CASE_REPO}/scripts/mark-tested.sh`, etc.)
 - [ ] Verify all three include the AGENT_RESULT output format section
 
 ## Validation Commands
diff --git a/docs/ideation/case-multi-agent/spec-phase-3.md b/docs/ideation/case-multi-agent/spec-phase-3.md
index a3ef360..143dca7 100644
--- a/docs/ideation/case-multi-agent/spec-phase-3.md
+++ b/docs/ideation/case-multi-agent/spec-phase-3.md
@@ -23,13 +23,13 @@ The orchestrator uses the Claude Code `Agent` tool to spawn subagents, passing e
 
 ### Orchestrator SKILL.md Rewrite
 
-**Pattern to follow**: Current `skills/case/SKILL.md` (preserve argument parsing, routing table, rules sections). Ideation's `execute-spec/SKILL.md` (for subagent dispatch pattern at `/Users/nicknisi/.claude/plugins/cache/nicknisi/ideation/0.9.0/skills/execute-spec/`).
+**Pattern to follow**: Current `skills/case/SKILL.md` (preserve argument parsing, routing table, rules sections). Ideation's `execute-spec/SKILL.md` (for subagent dispatch pattern in the ideation plugin's `skills/execute-spec/` directory).
 
 **Overview**: The SKILL.md becomes an orchestrator document. It retains existing sections (Arguments, Rules, Always Load, Task Routing, Project Manifest, Verification Tools) but replaces the monolithic execution flow with a phased subagent dispatch.
 
 **Key decisions**:
 
-- **Agent prompt paths are absolute**: The orchestrator reads agent prompt files from `/Users/nicknisi/Developer/case/agents/` (the `agents/` directory at the case repo root, NOT relative to CWD). SKILL.md must use the full absolute path since `/case` runs from target repos, not from the case repo. This matches how SKILL.md already references scripts with absolute paths.
+- **Agent prompt paths are absolute**: The orchestrator reads agent prompt files from `${CASE_REPO}/agents/` (the `agents/` directory at the case repo root, NOT relative to CWD). SKILL.md must use the full absolute path since `/case` runs from target repos, not from the case repo. This matches how SKILL.md already references scripts with absolute paths.
 - Each subagent is spawned sequentially (implementer → verifier → closer). No parallelism within a single `/case` run — the pipeline is linear.
 - The orchestrator creates both the `.md` task file AND the `.task.json` companion at the same time.
 - Baseline smoke test uses `scripts/bootstrap.sh` which already exists and validates repo readiness.
@@ -71,11 +71,11 @@ ORCHESTRATOR FLOW
 
 0. CHECK FOR EXISTING TASK (idempotent re-entry)
    - Read .case-active — if it contains a task ID, look up that specific
-     /Users/nicknisi/Developer/case/tasks/active/{task-id}.task.json
+     ${CASE_REPO}/tasks/active/{task-id}.task.json
      directly (fastest, most reliable)
    - If .case-active missing or empty: derive repo name from git remote,
      then match by argument type:
-       GitHub issue → scan /Users/nicknisi/Developer/case/tasks/active/*.task.json
+       GitHub issue → scan ${CASE_REPO}/tasks/active/*.task.json
                       for matching repo + issueType "github" + issue number
        Linear ID    → scan for matching issueType "linear" + issue ID
        Free text    → no automatic re-entry (ambiguous). Proceed to step 1.
@@ -99,8 +99,8 @@ ORCHESTRATOR FLOW
 
 2. TASK SETUP (updated for new format)
    - Find next sequential task number
-   - Create task file (.md) in /Users/nicknisi/Developer/case/tasks/active/ using appropriate template
-   - Create companion .task.json in /Users/nicknisi/Developer/case/tasks/active/ with initial values:
+   - Create task file (.md) in ${CASE_REPO}/tasks/active/ using appropriate template
+   - Create companion .task.json in ${CASE_REPO}/tasks/active/ with initial values:
      {
        "id": "<repo>-<n>-issue-<number>",
        "status": "active",
@@ -132,7 +132,7 @@ ORCHESTRATOR FLOW
      If exists: git checkout <branch>  (resume)
      If not: git checkout -b <branch>  (create)
    - Run baseline smoke test:
-     bash /Users/nicknisi/Developer/case/scripts/bootstrap.sh <repo-name>
+     bash ${CASE_REPO}/scripts/bootstrap.sh <repo-name>
    - If FAIL: stop, report broken baseline, suggest fixing before proceeding
    - If PASS: continue
    - Append to progress log:
@@ -143,7 +143,7 @@ ORCHESTRATOR FLOW
    - Update task JSON: orchestrator status → completed
 
 4. SPAWN IMPLEMENTER
-   - Read /Users/nicknisi/Developer/case/agents/implementer.md
+   - Read ${CASE_REPO}/agents/implementer.md
    - Use Agent tool:
      - prompt: <implementer.md content> + task context (file path, repo path,
        issue summary, playbook)
@@ -154,7 +154,7 @@ ORCHESTRATOR FLOW
    - If status == "completed": continue
 
 5. SPAWN VERIFIER
-   - Read /Users/nicknisi/Developer/case/agents/verifier.md
+   - Read ${CASE_REPO}/agents/verifier.md
    - Use Agent tool:
      - prompt: <verifier.md content> + task context (file path, repo path)
      - subagent_type: general-purpose
@@ -179,8 +179,8 @@ ORCHESTRATOR FLOW
 
 6. SPAWN CLOSER
    - Update task JSON: status → closing
-     bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> status closing
-   - Read /Users/nicknisi/Developer/case/agents/closer.md
+     bash ${CASE_REPO}/scripts/task-status.sh <task.json> status closing
+   - Read ${CASE_REPO}/agents/closer.md
    - Use Agent tool:
      - prompt: <closer.md content> + task context (file path, repo path,
        verifier AGENT_RESULT)
@@ -214,7 +214,7 @@ ORCHESTRATOR FLOW
 
 **Sections to add**:
 
-- `## Agent Architecture` — brief overview of the four-agent pipeline, reference to agents/ directory at `/Users/nicknisi/Developer/case/agents/`
+- `## Agent Architecture` — brief overview of the four-agent pipeline, reference to agents/ directory at `${CASE_REPO}/agents/`
 - `## Subagent Output Contract` — the AGENT_RESULT JSON schema that all subagents must produce
 - `## Baseline Smoke Test` — explain the bootstrap.sh check and what to do if it fails
 - `## Re-entry Semantics` — how the orchestrator resumes from `.task.json` state if a prior run was interrupted
@@ -247,7 +247,7 @@ ORCHESTRATOR FLOW
 | Verifier fails + src/ changed | Verification is mandatory (hook blocks without it). Options: fix-and-retry or abort. No skip. |
 | Verifier fails + no src/ changed | Verification is optional. Options: fix-and-retry, skip, or abort. |
 | Closer fails (hooks block PR) | Report missing prerequisites. Suggest which steps to run (mark-tested.sh, mark-manual-tested.sh). User decides next step. |
-| Agent prompt file not found | Stop with clear error: "Agent prompt file not found at /Users/nicknisi/Developer/case/agents/{name}.md — is the case plugin installed correctly?" |
+| Agent prompt file not found | Stop with clear error: "Agent prompt file not found at ${CASE_REPO}/agents/{name}.md — is the case plugin installed correctly?" |
 | No AGENT_RESULT in subagent response | Treat as failed. Log the raw response for debugging. Report to user. |
 | Re-entry: existing task found | Resume from last completed phase. Do not recreate task file or branch. |
 | Task JSON companion missing | Orchestrator creates it. If it goes missing mid-flow, recreate with current known state. |
diff --git a/docs/ideation/case-video-verification/spec.md b/docs/ideation/case-video-verification/spec.md
index 9faf9af..d458794 100644
--- a/docs/ideation/case-video-verification/spec.md
+++ b/docs/ideation/case-video-verification/spec.md
@@ -38,7 +38,7 @@ Wire `playwright-cli video-start` / `video-stop` into the verifier agent's workf
    ```
 4. Add video upload after screenshots:
    ```bash
-   VIDEO=$(/Users/nicknisi/Developer/case/scripts/upload-screenshot.sh /tmp/verification.webm)
+   VIDEO=$(${CASE_REPO}/scripts/upload-screenshot.sh /tmp/verification.webm)
    echo "$VIDEO"
    ```
    The upload script returns `<video src="url" controls></video>` for video files.
diff --git a/docs/ideation/context-engineering/spec-phase-3.md b/docs/ideation/context-engineering/spec-phase-3.md
index 78c1bd3..5bef1ce 100644
--- a/docs/ideation/context-engineering/spec-phase-3.md
+++ b/docs/ideation/context-engineering/spec-phase-3.md
@@ -166,7 +166,7 @@ After updating learnings, scan the learnings file for patterns:
 **What to add** (after step 5 "Read projects.json"):
 
 ```markdown
-6. Read `/Users/nicknisi/Developer/case/docs/learnings/{repo}.md` for tactical knowledge from previous tasks in this repo
+6. Read `${CASE_REPO}/docs/learnings/{repo}.md` for tactical knowledge from previous tasks in this repo
 ```
 
 **Implementation steps**:
diff --git a/docs/ideation/harness-improvements/spec-phase-1.md b/docs/ideation/harness-improvements/spec-phase-1.md
index c44d0a8..22f8bc8 100644
--- a/docs/ideation/harness-improvements/spec-phase-1.md
+++ b/docs/ideation/harness-improvements/spec-phase-1.md
@@ -131,9 +131,9 @@ Update the test piping command in `agents/implementer.md` to recommend JSON repo
 ```markdown
 # In the Validate section, add guidance:
 # Prefer JSON reporter for structured evidence:
-pnpm test --reporter=json 2>&1 | bash /Users/nicknisi/Developer/case/scripts/mark-tested.sh
+pnpm test --reporter=json 2>&1 | bash ${CASE_REPO}/scripts/mark-tested.sh
 # Fallback if JSON reporter unavailable:
-pnpm test 2>&1 | bash /Users/nicknisi/Developer/case/scripts/mark-tested.sh
+pnpm test 2>&1 | bash ${CASE_REPO}/scripts/mark-tested.sh
 ```
 
 ## Testing Requirements
diff --git a/docs/ideation/harness-improvements/spec-phase-2.md b/docs/ideation/harness-improvements/spec-phase-2.md
index b29f564..045b806 100644
--- a/docs/ideation/harness-improvements/spec-phase-2.md
+++ b/docs/ideation/harness-improvements/spec-phase-2.md
@@ -11,7 +11,7 @@ The Anthropic article identified this as the single highest-leverage interventio
 
 ## Feedback Strategy
 
-**Inner-loop command**: `bash scripts/session-start.sh /Users/nicknisi/Developer/case`
+**Inner-loop command**: `bash scripts/session-start.sh ${CASE_REPO}`
 
 **Playground**: The script itself — run it against the case repo and inspect output.
 
@@ -129,7 +129,7 @@ For each agent file, insert before the existing first numbered step:
 
 Run the session-start script to orient yourself:
 ```bash
-SESSION=$(bash /Users/nicknisi/Developer/case/scripts/session-start.sh <target-repo-path> --task <task.json>)
+SESSION=$(bash ${CASE_REPO}/scripts/session-start.sh <target-repo-path> --task <task.json>)
 echo "$SESSION"
 ```
 
@@ -166,13 +166,13 @@ Read the output to understand: current branch, last commits, task status, which
 bash -n scripts/session-start.sh
 
 # Run against case repo
-bash scripts/session-start.sh /Users/nicknisi/Developer/case
+bash scripts/session-start.sh ${CASE_REPO}
 
 # Validate JSON output
 bash scripts/session-start.sh . | node -e "const d=JSON.parse(require('fs').readFileSync('/dev/stdin','utf8')); console.log('branch:', d.repo.branch); console.log('Valid JSON: true')"
 
 # Run with task
-bash scripts/session-start.sh . --task /Users/nicknisi/Developer/case/tasks/active/authkit-nextjs-1-issue-364-proxy-support.task.json
+bash scripts/session-start.sh . --task ${CASE_REPO}/tasks/active/authkit-nextjs-1-issue-364-proxy-support.task.json
 
 # Verify agent files were updated
 for f in agents/implementer.md agents/verifier.md agents/closer.md agents/retrospective.md; do
diff --git a/docs/ideation/harness-improvements/spec-phase-3.md b/docs/ideation/harness-improvements/spec-phase-3.md
index 4eb5937..4447ce4 100644
--- a/docs/ideation/harness-improvements/spec-phase-3.md
+++ b/docs/ideation/harness-improvements/spec-phase-3.md
@@ -141,7 +141,7 @@ EOF
 echo ".case-reviewed created (${WARNINGS} warnings, ${INFO} info)" >&2
 
 # Update task JSON if .case-active present
-CASE_REPO="/Users/nicknisi/Developer/case"
+CASE_REPO="${CASE_REPO}"
 if [[ -f ".case-active" ]]; then
   TASK_ID=$(cat .case-active | tr -d '[:space:]')
   TASK_JSON="${CASE_REPO}/tasks/active/${TASK_ID}.task.json"
@@ -180,7 +180,7 @@ Insert after the manual testing check block (after the `fi` closing Check 3):
 # Check 5: Code review evidence (.case-reviewed marker)
 if [[ ! -f ".case-reviewed" ]]; then
   FAILURES+=("[FAIL] Code review not done — .case-reviewed marker missing")
-  FIXES+=("  FIX: Run the reviewer agent, then: bash /Users/nicknisi/Developer/case/scripts/mark-reviewed.sh --critical 0 --warnings N --info N")
+  FIXES+=("  FIX: Run the reviewer agent, then: bash ${CASE_REPO}/scripts/mark-reviewed.sh --critical 0 --warnings N --info N")
 elif ! grep -q "critical: 0" ".case-reviewed" 2>/dev/null; then
   FAILURES+=("[FAIL] Code review has unresolved critical findings")
   FIXES+=("  FIX: Address critical findings from the reviewer, then re-run the reviewer agent")
diff --git a/docs/learnings/README.md b/docs/learnings/README.md
index d47d6db..335919c 100644
--- a/docs/learnings/README.md
+++ b/docs/learnings/README.md
@@ -1,22 +1,40 @@
-# Repo Learnings
+# Repo Learnings (External)
 
-Tactical knowledge accumulated by the retrospective agent across pipeline runs. Each file corresponds to a repo in `projects.json`.
+Per-repo tactical knowledge is stored in a separate GitHub repo, configured via the `CASE_LEARNINGS_REPO` environment variable. This keeps codebase-specific knowledge out of the open-source harness repo.
+
+## Setup
+
+1. Create a GitHub repo for learnings:
+   ```bash
+   gh repo create case-learnings --public --description "Per-repo tactical knowledge for case harness"
+   ```
+
+2. Set the env var:
+   ```bash
+   export CASE_LEARNINGS_REPO='youruser/case-learnings'
+   ```
+   Add this to your shell profile or Claude Code settings for persistence.
 
 ## How it works
 
-1. After every `/case` pipeline run, the retrospective agent analyzes what happened
-2. If it discovers tactical knowledge specific to a repo, it appends to that repo's learnings file
-3. The implementer agent reads the relevant learnings file during setup, before writing code
-4. If the same issue appears 3+ times in a learnings file, the retrospective escalates it to a convention or golden principle
+1. After every pipeline run, the retrospective agent calls `scripts/write-learning.sh` to append tactical knowledge
+2. Before coding, the implementer agent calls `scripts/read-learning.sh` to load tactical knowledge
+3. If 3+ similar entries accumulate, the retrospective escalates to a convention or golden principle (in the case repo)
+4. Each fork/user has their own learnings repo — knowledge doesn't leak across forks
 
 ## Format
 
-Each entry is a dated bullet point with context:
+Each entry is a dated bullet point:
 
 ```markdown
-- **2026-03-08** — `src/middleware.ts`: Mock `next/headers` as a module, not individual exports. Individual mocks cause type errors in strict mode. (from task authkit-nextjs-1-issue-53)
+- **2026-03-08** — `src/middleware.ts`: Mock `next/headers` as a module, not individual exports. (from task authkit-nextjs-1-issue-53)
 ```
 
+## Scripts
+
+- `scripts/read-learning.sh <repo>` — read a repo's learnings (stdout)
+- `scripts/write-learning.sh <repo> <entry>` — append an entry and commit to the external repo
+
 ## Rules
 
 - Agents append entries — never edit or remove existing ones
diff --git a/docs/learnings/authkit-nextjs.md b/docs/learnings/authkit-nextjs.md
deleted file mode 100644
index 0d63ad3..0000000
--- a/docs/learnings/authkit-nextjs.md
+++ /dev/null
@@ -1,5 +0,0 @@
-# AuthKit Next.js Learnings
-
-Tactical knowledge from completed tasks. Read by agents before working in this repo.
-
-<!-- Retrospective agent appends entries below. Do not edit existing entries. -->
diff --git a/docs/learnings/authkit-session.md b/docs/learnings/authkit-session.md
deleted file mode 100644
index 5af1849..0000000
--- a/docs/learnings/authkit-session.md
+++ /dev/null
@@ -1,5 +0,0 @@
-# AuthKit Session Learnings
-
-Tactical knowledge from completed tasks. Read by agents before working in this repo.
-
-<!-- Retrospective agent appends entries below. Do not edit existing entries. -->
diff --git a/docs/learnings/authkit-tanstack-start.md b/docs/learnings/authkit-tanstack-start.md
deleted file mode 100644
index 0dd9f96..0000000
--- a/docs/learnings/authkit-tanstack-start.md
+++ /dev/null
@@ -1,5 +0,0 @@
-# AuthKit TanStack Start Learnings
-
-Tactical knowledge from completed tasks. Read by agents before working in this repo.
-
-<!-- Retrospective agent appends entries below. Do not edit existing entries. -->
diff --git a/docs/learnings/cli.md b/docs/learnings/cli.md
deleted file mode 100644
index 694d223..0000000
--- a/docs/learnings/cli.md
+++ /dev/null
@@ -1,5 +0,0 @@
-# CLI Learnings
-
-Tactical knowledge from completed tasks. Read by agents before working in this repo.
-
-<!-- Retrospective agent appends entries below. Do not edit existing entries. -->
diff --git a/docs/learnings/skills.md b/docs/learnings/skills.md
deleted file mode 100644
index 9000b6d..0000000
--- a/docs/learnings/skills.md
+++ /dev/null
@@ -1,5 +0,0 @@
-# Skills Learnings
-
-Tactical knowledge from completed tasks. Read by agents before working in this repo.
-
-<!-- Retrospective agent appends entries below. Do not edit existing entries. -->
diff --git a/hooks/post-pr-cleanup.sh b/hooks/post-pr-cleanup.sh
index a4934cf..016c687 100755
--- a/hooks/post-pr-cleanup.sh
+++ b/hooks/post-pr-cleanup.sh
@@ -7,7 +7,8 @@
 
 set -euo pipefail
 
-CASE_REPO="/Users/nicknisi/Developer/case"
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+CASE_REPO="$(cd "$SCRIPT_DIR/.." && pwd)"
 
 # Read hook input from stdin
 INPUT=$(cat)
diff --git a/hooks/pre-pr-check.sh b/hooks/pre-pr-check.sh
index 1faa668..3b24d73 100755
--- a/hooks/pre-pr-check.sh
+++ b/hooks/pre-pr-check.sh
@@ -5,6 +5,9 @@
 
 set -uo pipefail
 
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+CASE_REPO="$(cd "$SCRIPT_DIR/.." && pwd)"
+
 # Read hook input from stdin
 INPUT=$(cat)
 
@@ -44,10 +47,10 @@ fi
 # Check 2: Tests were run (.case-tested marker with evidence)
 if [[ ! -f ".case-tested" ]]; then
   FAILURES+=("[FAIL] Tests not verified — .case-tested marker missing")
-  FIXES+=("  FIX: Run tests and pipe output: pnpm test 2>&1 | bash /Users/nicknisi/Developer/case/scripts/mark-tested.sh")
+  FIXES+=("  FIX: Run tests and pipe output: pnpm test 2>&1 | bash ${CASE_REPO}/scripts/mark-tested.sh")
 elif ! grep -q "output_hash:" ".case-tested" 2>/dev/null; then
   FAILURES+=("[FAIL] Tests marker has no evidence — was created with 'touch' instead of the mark-tested script")
-  FIXES+=("  FIX: Run tests properly: pnpm test 2>&1 | bash /Users/nicknisi/Developer/case/scripts/mark-tested.sh")
+  FIXES+=("  FIX: Run tests properly: pnpm test 2>&1 | bash ${CASE_REPO}/scripts/mark-tested.sh")
 fi
 
 # Check 3: Manual testing (smart — only required if src/ files changed)
@@ -63,10 +66,10 @@ fi
 if [[ "$NEEDS_MANUAL_TEST" == "true" ]]; then
   if [[ ! -f ".case-manual-tested" ]]; then
     FAILURES+=("[FAIL] Manual testing not done — .case-manual-tested marker missing")
-    FIXES+=("  FIX: Test in the example app with playwright-cli, then: bash /Users/nicknisi/Developer/case/scripts/mark-manual-tested.sh")
+    FIXES+=("  FIX: Test in the example app with playwright-cli, then: bash ${CASE_REPO}/scripts/mark-manual-tested.sh")
   elif ! grep -q "evidence:" ".case-manual-tested" 2>/dev/null; then
     FAILURES+=("[FAIL] Manual testing marker has no evidence — was created with 'touch' instead of the mark script")
-    FIXES+=("  FIX: Use playwright-cli to test, then: bash /Users/nicknisi/Developer/case/scripts/mark-manual-tested.sh")
+    FIXES+=("  FIX: Use playwright-cli to test, then: bash ${CASE_REPO}/scripts/mark-manual-tested.sh")
   fi
 fi
 
@@ -98,7 +101,7 @@ fi
 # Check 5: Code review evidence (.case-reviewed marker)
 if [[ ! -f ".case-reviewed" ]]; then
   FAILURES+=("[FAIL] Code review not done — .case-reviewed marker missing")
-  FIXES+=("  FIX: Run the reviewer agent, then: bash /Users/nicknisi/Developer/case/scripts/mark-reviewed.sh --critical 0 --warnings N --info N")
+  FIXES+=("  FIX: Run the reviewer agent, then: bash ${CASE_REPO}/scripts/mark-reviewed.sh --critical 0 --warnings N --info N")
 elif ! grep -q "critical: 0" ".case-reviewed" 2>/dev/null; then
   FAILURES+=("[FAIL] Code review has unresolved critical findings")
   FIXES+=("  FIX: Address critical findings from the reviewer, then re-run the reviewer agent")
diff --git a/scripts/mark-manual-tested.sh b/scripts/mark-manual-tested.sh
index 1cb7f23..1516252 100755
--- a/scripts/mark-manual-tested.sh
+++ b/scripts/mark-manual-tested.sh
@@ -54,7 +54,8 @@ EOF
 echo ".case-manual-tested created (${EVIDENCE_DETAILS})" >&2
 
 # Update task JSON if .case-active contains a task ID
-CASE_REPO="/Users/nicknisi/Developer/case"
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+CASE_REPO="$(cd "$SCRIPT_DIR/.." && pwd)"
 if [[ -f ".case-active" ]]; then
   TASK_ID=$(cat .case-active | tr -d '[:space:]')
   TASK_JSON="${CASE_REPO}/tasks/active/${TASK_ID}.task.json"
diff --git a/scripts/mark-reviewed.sh b/scripts/mark-reviewed.sh
index aa4e806..9588c4a 100755
--- a/scripts/mark-reviewed.sh
+++ b/scripts/mark-reviewed.sh
@@ -38,7 +38,8 @@ EOF
 echo ".case-reviewed created (${WARNINGS} warnings, ${INFO} info)" >&2
 
 # Update task JSON if .case-active present
-CASE_REPO="/Users/nicknisi/Developer/case"
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+CASE_REPO="$(cd "$SCRIPT_DIR/.." && pwd)"
 if [[ -f ".case-active" ]]; then
   TASK_ID=$(cat .case-active | tr -d '[:space:]')
   TASK_JSON="${CASE_REPO}/tasks/active/${TASK_ID}.task.json"
diff --git a/scripts/mark-tested.sh b/scripts/mark-tested.sh
index b236fae..26c981a 100755
--- a/scripts/mark-tested.sh
+++ b/scripts/mark-tested.sh
@@ -67,7 +67,7 @@ fi
 echo ".case-tested created (hash: ${OUTPUT_HASH:0:12}...)" >&2
 
 # Update task JSON if .case-active contains a task ID
-CASE_REPO="/Users/nicknisi/Developer/case"
+CASE_REPO="$(cd "$SCRIPT_DIR/.." && pwd)"
 if [[ -f ".case-active" ]]; then
   TASK_ID=$(cat .case-active | tr -d '[:space:]')
   TASK_JSON="${CASE_REPO}/tasks/active/${TASK_ID}.task.json"
diff --git a/scripts/read-learning.sh b/scripts/read-learning.sh
new file mode 100755
index 0000000..a293c90
--- /dev/null
+++ b/scripts/read-learning.sh
@@ -0,0 +1,44 @@
+#!/usr/bin/env bash
+# Read a repo's learnings file from the external learnings repo.
+# Outputs the file content to stdout. If no file exists, outputs nothing.
+#
+# Usage: read-learning.sh <repo-name>
+#
+# Requires:
+#   - gh CLI authenticated
+#   - CASE_LEARNINGS_REPO env var set (e.g., 'youruser/case-learnings')
+
+set -euo pipefail
+
+if [[ -z "${CASE_LEARNINGS_REPO:-}" ]]; then
+  echo "ERROR: CASE_LEARNINGS_REPO is not set." >&2
+  echo "" >&2
+  echo "This script reads per-repo learnings from a GitHub repo." >&2
+  echo "Set CASE_LEARNINGS_REPO to your own GitHub repo (e.g., 'youruser/case-learnings')." >&2
+  echo "" >&2
+  echo "Setup:" >&2
+  echo "  1. Create a GitHub repo for learnings (e.g., gh repo create case-learnings --public)" >&2
+  echo "  2. Export the env var: export CASE_LEARNINGS_REPO='youruser/case-learnings'" >&2
+  echo "  3. Or add it to your shell profile / Claude Code settings" >&2
+  exit 1
+fi
+
+if [[ $# -lt 1 ]]; then
+  echo "Usage: read-learning.sh <repo-name>" >&2
+  exit 1
+fi
+
+REPO_NAME="$1"
+FILE_PATH="${REPO_NAME}.md"
+
+# Fetch file content via GitHub API (base64-encoded).
+# If the file doesn't exist (404), output nothing.
+CONTENT=$(gh api "repos/${CASE_LEARNINGS_REPO}/contents/${FILE_PATH}" --jq '.content' 2>/dev/null) || true
+
+if [[ -z "$CONTENT" ]]; then
+  echo "No learnings file found for ${REPO_NAME} in ${CASE_LEARNINGS_REPO}." >&2
+  exit 0
+fi
+
+# Decode base64 (macOS uses --decode, Linux uses -d)
+echo "$CONTENT" | base64 --decode 2>/dev/null || echo "$CONTENT" | base64 -d
diff --git a/scripts/upload-screenshot.sh b/scripts/upload-screenshot.sh
index 5e65f7d..4e69aaf 100755
--- a/scripts/upload-screenshot.sh
+++ b/scripts/upload-screenshot.sh
@@ -18,7 +18,20 @@
 
 set -euo pipefail
 
-ASSETS_REPO="nicknisi/case-assets"
+if [[ -z "${CASE_ASSETS_REPO:-}" ]]; then
+  echo "ERROR: CASE_ASSETS_REPO is not set." >&2
+  echo "" >&2
+  echo "This script uploads screenshots/videos to a GitHub repo as release assets." >&2
+  echo "Set CASE_ASSETS_REPO to your own GitHub repo (e.g., 'youruser/case-assets')." >&2
+  echo "" >&2
+  echo "Setup:" >&2
+  echo "  1. Create a GitHub repo for assets (e.g., gh repo create case-assets --public)" >&2
+  echo "  2. Export the env var: export CASE_ASSETS_REPO='youruser/case-assets'" >&2
+  echo "  3. Or add it to your shell profile / Claude Code settings" >&2
+  exit 1
+fi
+
+ASSETS_REPO="$CASE_ASSETS_REPO"
 RELEASE_TAG="assets"
 
 if [[ $# -lt 1 ]]; then
diff --git a/scripts/write-learning.sh b/scripts/write-learning.sh
new file mode 100755
index 0000000..4d82282
--- /dev/null
+++ b/scripts/write-learning.sh
@@ -0,0 +1,120 @@
+#!/usr/bin/env bash
+# Append a learning entry to a repo's learnings file in the external repo.
+# Creates the file with a standard header if it doesn't exist.
+#
+# Usage: write-learning.sh <repo-name> "<entry>"
+#    or: echo "<entry>" | write-learning.sh <repo-name>
+#
+# Requires:
+#   - gh CLI authenticated
+#   - CASE_LEARNINGS_REPO env var set (e.g., 'youruser/case-learnings')
+
+set -euo pipefail
+
+if [[ -z "${CASE_LEARNINGS_REPO:-}" ]]; then
+  echo "ERROR: CASE_LEARNINGS_REPO is not set." >&2
+  echo "" >&2
+  echo "This script writes per-repo learnings to a GitHub repo." >&2
+  echo "Set CASE_LEARNINGS_REPO to your own GitHub repo (e.g., 'youruser/case-learnings')." >&2
+  echo "" >&2
+  echo "Setup:" >&2
+  echo "  1. Create a GitHub repo for learnings (e.g., gh repo create case-learnings --public)" >&2
+  echo "  2. Export the env var: export CASE_LEARNINGS_REPO='youruser/case-learnings'" >&2
+  echo "  3. Or add it to your shell profile / Claude Code settings" >&2
+  exit 1
+fi
+
+if [[ $# -lt 1 ]]; then
+  echo "Usage: write-learning.sh <repo-name> [entry]" >&2
+  exit 1
+fi
+
+REPO_NAME="$1"
+ENTRY="${2:-}"
+
+# If no entry argument, read from stdin
+if [[ -z "$ENTRY" ]]; then
+  ENTRY=$(cat)
+fi
+
+if [[ -z "$ENTRY" ]]; then
+  echo "ERROR: No entry provided. Pass as argument or pipe via stdin." >&2
+  exit 1
+fi
+
+FILE_PATH="${REPO_NAME}.md"
+
+# Fetch existing file for SHA (required by GitHub API for updates) and content
+EXISTING_SHA=""
+EXISTING_CONTENT=""
+
+RESPONSE=$(gh api "repos/${CASE_LEARNINGS_REPO}/contents/${FILE_PATH}" 2>/dev/null) || true
+
+if [[ -n "$RESPONSE" ]]; then
+  EXISTING_SHA=$(echo "$RESPONSE" | gh api --input - --jq '.sha' 2>/dev/null || echo "$RESPONSE" | node -e "process.stdin.on('data',d=>{console.log(JSON.parse(d).sha||'')})" 2>/dev/null) || true
+  ENCODED_CONTENT=$(echo "$RESPONSE" | gh api --input - --jq '.content' 2>/dev/null || echo "$RESPONSE" | node -e "process.stdin.on('data',d=>{console.log(JSON.parse(d).content||'')})" 2>/dev/null) || true
+  if [[ -n "$ENCODED_CONTENT" ]]; then
+    EXISTING_CONTENT=$(echo "$ENCODED_CONTENT" | base64 --decode 2>/dev/null || echo "$ENCODED_CONTENT" | base64 -d 2>/dev/null) || true
+  fi
+fi
+
+# Build new content
+if [[ -z "$EXISTING_CONTENT" ]]; then
+  # Create new file with standard header
+  DISPLAY_NAME=$(echo "$REPO_NAME" | sed 's/-/ /g' | awk '{for(i=1;i<=NF;i++) $i=toupper(substr($i,1,1)) substr($i,2)}1')
+  NEW_CONTENT="# ${DISPLAY_NAME} Learnings
+
+Tactical knowledge from completed tasks. Read by agents before working in this repo.
+
+<!-- Retrospective agent appends entries below. Do not edit existing entries. -->
+${ENTRY}
+"
+else
+  NEW_CONTENT="${EXISTING_CONTENT}${ENTRY}
+"
+fi
+
+# Base64 encode the new content
+ENCODED=$(printf '%s' "$NEW_CONTENT" | base64)
+
+COMMIT_MSG="learnings(${REPO_NAME}): append entry"
+
+# Write with retry for SHA conflicts (max 2 attempts)
+MAX_RETRIES=2
+ATTEMPT=0
+
+while [[ $ATTEMPT -lt $MAX_RETRIES ]]; do
+  ATTEMPT=$((ATTEMPT + 1))
+
+  if [[ -n "$EXISTING_SHA" ]]; then
+    # Update existing file
+    if gh api "repos/${CASE_LEARNINGS_REPO}/contents/${FILE_PATH}" \
+      --method PUT \
+      -f message="$COMMIT_MSG" \
+      -f content="$ENCODED" \
+      -f sha="$EXISTING_SHA" \
+      > /dev/null 2>&1; then
+      echo "Appended learning to ${CASE_LEARNINGS_REPO}/${FILE_PATH}" >&2
+      exit 0
+    fi
+  else
+    # Create new file
+    if gh api "repos/${CASE_LEARNINGS_REPO}/contents/${FILE_PATH}" \
+      --method PUT \
+      -f message="$COMMIT_MSG" \
+      -f content="$ENCODED" \
+      > /dev/null 2>&1; then
+      echo "Created ${CASE_LEARNINGS_REPO}/${FILE_PATH} with learning entry" >&2
+      exit 0
+    fi
+  fi
+
+  # On failure, refresh SHA and retry
+  if [[ $ATTEMPT -lt $MAX_RETRIES ]]; then
+    echo "Write conflict, retrying (attempt $((ATTEMPT + 1))/${MAX_RETRIES})..." >&2
+    EXISTING_SHA=$(gh api "repos/${CASE_LEARNINGS_REPO}/contents/${FILE_PATH}" --jq '.sha' 2>/dev/null) || true
+  fi
+done
+
+echo "ERROR: Failed to write learning after ${MAX_RETRIES} attempts." >&2
+exit 1
diff --git a/skills/case/SKILL.md b/skills/case/SKILL.md
index 777c614..c3f4164 100644
--- a/skills/case/SKILL.md
+++ b/skills/case/SKILL.md
@@ -9,9 +9,15 @@ argument-hint: "[issue-number] or [LINEAR-ID]"
 You are operating within the Case harness for WorkOS open source projects.
 Humans steer. Agents execute. When agents struggle, fix the harness.
 
-**Case repo**: `/Users/nicknisi/Developer/case`
+**Case repo**: Resolve dynamically at the start of each `/case` invocation. This SKILL.md file lives at `skills/case/SKILL.md` within the case repo. Strip that suffix from this file's absolute path to get `CASE_REPO`. All `${CASE_REPO}` references below use this resolved path.
 
-All paths below are relative to the skill's cache directory. For scripts, tasks, and project manifest, use the case repo path above.
+Before running any scripts, resolve and verify the case repo path:
+```bash
+# CASE_REPO is the root of the case harness repo (parent of skills/, agents/, scripts/)
+# Derive it from this skill file's location: strip skills/case/ from the path
+CASE_REPO="<resolved-path>"  # Replace with actual path derived from this SKILL.md's location
+ls "${CASE_REPO}/AGENTS.md" "${CASE_REPO}/projects.json" > /dev/null
+```
 
 ## Agent Architecture
 
@@ -26,7 +32,7 @@ Case uses a **six-agent pipeline** to prevent context pollution and enable self-
 | **Closer** | Create PR with thorough description, satisfy hook gates, post review comments | Read, Bash, Glob, Grep |
 | **Retrospective** | Apply harness improvements directly, maintain per-repo learnings | Read, Edit, Write, Bash, Glob, Grep |
 
-Agent prompt files: `/Users/nicknisi/Developer/case/agents/{implementer,verifier,reviewer,closer,retrospective}.md`
+Agent prompt files: `${CASE_REPO}/agents/{implementer,verifier,reviewer,closer,retrospective}.md`
 
 The orchestrator spawns each agent sequentially using the `Agent` tool, passing the agent prompt file content as the prompt. Each agent ends its response with a structured `AGENT_RESULT` block (see below).
 
@@ -91,7 +97,7 @@ After parsing and fetching the issue, execute this pipeline:
 **Explicit arguments always win over `.case-active`.** The `.case-active` shortcut is only used for no-arg `/case`. When an explicit issue number or Linear ID is provided, re-entry is matched by issue content, not by `.case-active`.
 
 1. **If an explicit argument was provided** (issue number or Linear ID):
-   - Scan `/Users/nicknisi/Developer/case/tasks/active/*.task.json` for a match:
+   - Scan `${CASE_REPO}/tasks/active/*.task.json` for a match:
      - GitHub issue → matching `repo` + `issueType: "github"` + `issue` number
      - Linear ID → matching `issueType: "linear"` + `issue` ID
    - Ignore `.case-active` — it may be stale or from a different issue
@@ -99,7 +105,7 @@ After parsing and fetching the issue, execute this pipeline:
    - If not found → proceed to step 1 (new task)
 
 2. **If no argument** (`/case` with no args):
-   - Read `.case-active` — if it contains a task ID, look up `/Users/nicknisi/Developer/case/tasks/active/{task-id}.task.json` directly
+   - Read `.case-active` — if it contains a task ID, look up `${CASE_REPO}/tasks/active/{task-id}.task.json` directly
    - If found → resume
    - If not found → load harness context (no orchestrator flow)
 
@@ -122,9 +128,9 @@ _(Already done in the Arguments section above)_
 ### Step 2: Task Setup
 
 1. Derive repo name from `git remote get-url origin`
-2. Find next sequential task number: count existing `{repo}-*.md` files in `/Users/nicknisi/Developer/case/tasks/active/` + 1
-3. Create task file (`.md`) using the appropriate template from `/Users/nicknisi/Developer/case/tasks/templates/`
-4. Create companion `.task.json` in `/Users/nicknisi/Developer/case/tasks/active/`:
+2. Find next sequential task number: count existing `{repo}-*.md` files in `${CASE_REPO}/tasks/active/` + 1
+3. Create task file (`.md`) using the appropriate template from `${CASE_REPO}/tasks/templates/`
+4. Create companion `.task.json` in `${CASE_REPO}/tasks/active/`:
    ```json
    {
      "id": "<repo>-<n>-issue-<number>",
@@ -163,7 +169,7 @@ _(Already done in the Arguments section above)_
    - Not exists → `git checkout -b <branch>` (create)
 3. Run baseline smoke test:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/bootstrap.sh <repo-name>
+   bash ${CASE_REPO}/scripts/bootstrap.sh <repo-name>
    ```
    - If FAIL: Report broken baseline to user via `AskUserQuestion`. Do not spawn implementer. **Go to step 9 (Retrospective)** with outcome "failed" and failed agent "orchestrator/baseline".
    - If PASS: continue
@@ -176,13 +182,13 @@ _(Already done in the Arguments section above)_
    ```
 5. Update task JSON:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent orchestrator status completed
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> agent orchestrator completed now
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent orchestrator status completed
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> agent orchestrator completed now
    ```
 
 ### Step 4: Spawn Implementer
 
-1. Read `/Users/nicknisi/Developer/case/agents/implementer.md`
+1. Read `${CASE_REPO}/agents/implementer.md`
 2. Use the `Agent` tool:
    - **prompt**: `<implementer.md content>` + task context:
      - Task file path (`.md` and `.task.json`)
@@ -197,7 +203,7 @@ _(Already done in the Arguments section above)_
 
 ### Step 5: Spawn Verifier
 
-1. Read `/Users/nicknisi/Developer/case/agents/verifier.md`
+1. Read `${CASE_REPO}/agents/verifier.md`
 2. Use the `Agent` tool:
    - **prompt**: `<verifier.md content>` + task context (file path, repo path)
    - **subagent_type**: `general-purpose`
@@ -217,7 +223,7 @@ _(Already done in the Arguments section above)_
 
 ### Step 6: Spawn Reviewer
 
-1. Read `/Users/nicknisi/Developer/case/agents/reviewer.md`
+1. Read `${CASE_REPO}/agents/reviewer.md`
 2. Use the `Agent` tool:
    - **prompt**: `<reviewer.md content>` + task context:
      - Task file path (`.md` and `.task.json`)
@@ -238,9 +244,9 @@ _(Already done in the Arguments section above)_
 
 1. Update task JSON status to `closing`:
    ```bash
-   bash /Users/nicknisi/Developer/case/scripts/task-status.sh <task.json> status closing
+   bash ${CASE_REPO}/scripts/task-status.sh <task.json> status closing
    ```
-2. Read `/Users/nicknisi/Developer/case/agents/closer.md`
+2. Read `${CASE_REPO}/agents/closer.md`
 3. Use the `Agent` tool:
    - **prompt**: `<closer.md content>` + task context (file path, repo path, verifier `AGENT_RESULT`, reviewer `AGENT_RESULT`)
    - **subagent_type**: `general-purpose`
@@ -259,7 +265,7 @@ Report to user:
 
 **Always runs** — after both successful and failed pipelines, at every failure class (baseline, implementer, verifier, reviewer, closer). Every failure branch in steps 3-8 routes here explicitly. This is how the harness improves itself.
 
-1. Read `/Users/nicknisi/Developer/case/agents/retrospective.md`
+1. Read `${CASE_REPO}/agents/retrospective.md`
 2. Use the `Agent` tool:
    - **prompt**: `<retrospective.md content>` + context:
      - Task file path (with progress log from all agents)
@@ -292,7 +298,7 @@ If `/case` is invoked and a `.task.json` already exists for the issue, the orche
 Before spawning the implementer, the orchestrator runs the target repo's test/build suite to confirm a clean baseline. This prevents agents from building on broken foundations.
 
 ```bash
-bash /Users/nicknisi/Developer/case/scripts/bootstrap.sh <repo-name>
+bash ${CASE_REPO}/scripts/bootstrap.sh <repo-name>
 ```
 
 The bootstrap script runs the repo's `setup`, `build`, `test`, `typecheck`, and `lint` commands (from `projects.json`). If any fail:
@@ -337,25 +343,25 @@ Based on the user's request, load the relevant context:
 
 ## Project Manifest
 
-Full repo metadata (paths, commands, remotes): `/Users/nicknisi/Developer/case/projects.json`
+Full repo metadata (paths, commands, remotes): `${CASE_REPO}/projects.json`
 
 ## Task Dispatch
 
 To create a task for async agent execution:
 
-1. Choose template from `/Users/nicknisi/Developer/case/tasks/templates/`
+1. Choose template from `${CASE_REPO}/tasks/templates/`
 2. Fill in `{placeholder}` fields
-3. Save `.md` to `/Users/nicknisi/Developer/case/tasks/active/{repo}-{n}-{slug}.md`
-4. Create companion `.task.json` with the same stem (see task schema: `/Users/nicknisi/Developer/case/tasks/task.schema.json`)
-5. Update task JSON status as agents complete work via `/Users/nicknisi/Developer/case/scripts/task-status.sh`
+3. Save `.md` to `${CASE_REPO}/tasks/active/{repo}-{n}-{slug}.md`
+4. Create companion `.task.json` with the same stem (see task schema: `${CASE_REPO}/tasks/task.schema.json`)
+5. Update task JSON status as agents complete work via `${CASE_REPO}/scripts/task-status.sh`
 
 Available templates:
-- `/Users/nicknisi/Developer/case/tasks/templates/cli-command.md` — add a CLI command
-- `/Users/nicknisi/Developer/case/tasks/templates/authkit-framework.md` — new AuthKit framework integration
-- `/Users/nicknisi/Developer/case/tasks/templates/bug-fix.md` — fix a bug in any repo
-- `/Users/nicknisi/Developer/case/tasks/templates/cross-repo-update.md` — coordinated cross-repo change
+- `${CASE_REPO}/tasks/templates/cli-command.md` — add a CLI command
+- `${CASE_REPO}/tasks/templates/authkit-framework.md` — new AuthKit framework integration
+- `${CASE_REPO}/tasks/templates/bug-fix.md` — fix a bug in any repo
+- `${CASE_REPO}/tasks/templates/cross-repo-update.md` — coordinated cross-repo change
 
-Format spec: `/Users/nicknisi/Developer/case/tasks/README.md`
+Format spec: `${CASE_REPO}/tasks/README.md`
 
 ## Working in a Target Repo
 
@@ -363,9 +369,9 @@ Before making changes in any target repo:
 
 1. Create a feature branch (or use `claude --worktree` for isolated work)
 2. Read that repo's `CLAUDE.md` or `CLAUDE.local.md` for project-specific instructions
-3. Run `/Users/nicknisi/Developer/case/scripts/bootstrap.sh {repo-name}` to verify readiness
+3. Run `${CASE_REPO}/scripts/bootstrap.sh {repo-name}` to verify readiness
 4. Follow the repo's PR checklist before opening a PR
-5. Run `/Users/nicknisi/Developer/case/scripts/check.sh --repo {repo-name}` to verify conventions
+5. Run `${CASE_REPO}/scripts/check.sh --repo {repo-name}` to verify conventions
 
 ## Verification Tools
 
@@ -438,8 +444,8 @@ cp .playwright-cli/page-*.png /tmp/after.png
 
 # Upload and get markdown
 # For video: the script auto-converts to GIF (inline) + mp4 (download link)
-VIDEO=$(/Users/nicknisi/Developer/case/scripts/upload-screenshot.sh /tmp/verification.webm)
-SCREENSHOT=$(/Users/nicknisi/Developer/case/scripts/upload-screenshot.sh /tmp/after.png)
+VIDEO=$(${CASE_REPO}/scripts/upload-screenshot.sh /tmp/verification.webm)
+SCREENSHOT=$(${CASE_REPO}/scripts/upload-screenshot.sh /tmp/after.png)
 
 # Use in PR body — VIDEO contains both GIF embed and mp4 download link
 echo "## Verification"
@@ -497,8 +503,8 @@ The orchestrator spawns the closer, which handles this checklist. If you're the
 
 When an agent struggles or produces poor output, the fix goes into case/, not the code:
 
-- Missing pattern? Add to `/Users/nicknisi/Developer/case/docs/architecture/`
-- Unclear convention? Update `/Users/nicknisi/Developer/case/docs/conventions/`
-- Recurring task? Add a playbook + template in `/Users/nicknisi/Developer/case/`
-- Agent violation? Add to `/Users/nicknisi/Developer/case/docs/golden-principles.md` and update `scripts/check.sh`
+- Missing pattern? Add to `${CASE_REPO}/docs/architecture/`
+- Unclear convention? Update `${CASE_REPO}/docs/conventions/`
+- Recurring task? Add a playbook + template in `${CASE_REPO}/`
+- Agent violation? Add to `${CASE_REPO}/docs/golden-principles.md` and update `scripts/check.sh`
 - Wrong approach? Update the relevant `CLAUDE.md` in the target repo
diff --git a/tasks/README.md b/tasks/README.md
index 659f210..853e553 100644
--- a/tasks/README.md
+++ b/tasks/README.md
@@ -52,7 +52,7 @@ The JSON file stores structured metadata that agents and scripts update programm
 
 Fields: `id`, `status`, `created`, `repo`, `issue`, `issueType`, `branch`, `agents`, `tested`, `manualTested`, `prUrl`, `prNumber`.
 
-Read/write via: `bash /Users/nicknisi/Developer/case/scripts/task-status.sh <file> <field> [value]`
+Read/write via: `bash ${CASE_REPO}/scripts/task-status.sh <file> <field> [value]`
 
 **Evidence flags** (`tested`, `manualTested`) can only be set by marker scripts (`mark-tested.sh`, `mark-manual-tested.sh`) — not by agents directly.