feat: add treehouse setup hook and manual PR validation flow

[michael-sb]

Intent

Captain, the developer wanted AGENTS.md updated so all firstmate instructions to crewmates for running the no-mistakes pipeline use the manual-PR workflow with --skip=pr,ci. They wanted validation to run through push without no-mistakes opening a GitHub PR, leaving the captain to open the PR manually afterward, while preserving the explicit rule that firstmate must not merge without approval. They asked for the change to be tightly scoped to the no-mistakes validation and PR-ready instructions, with a caveat that manual PR bodies may need to preserve the no-mistakes marker required by the current workflow checks. They also required creating branch fm/nomistakes-skip-pr-ci-v5, committing the change, staying inside the worktree, not pushing to origin or GitHub, and using only the local no-mistakes remote/bare gate if needed to start validation with push/pr/ci steps skipped.

What Changed

• Adds Treehouse post-create setup support in bootstrap/spawn, including hook installation, environment propagation, and a dedicated fm-treehouse-post-create.sh helper.
• Updates firstmate no-mistakes guidance and generated crewmate briefs so validation runs through branch push with --skip=pr,ci, leaving PR creation/manual merge approval outside the pipeline.
• Expands shell tests for bootstrap hook handling, generated brief wording, spawn Treehouse environment behavior, and post-create setup execution.

Risk Assessment

⚠️ Medium: Captain, no new material issues were found beyond the previously ignored Treehouse TOML edge case, but the branch changes bootstrap/spawn behavior and global Treehouse hook wiring with nontrivial shell parsing.

Testing

Captain, I inspected the branch diff and no-mistakes instructions, ran the focused changed-area tests, generated an actual crewmate brief as end-user evidence, verified it contains the manual-PR workflow and not the old push-skipping wording, ran the full shell test suite, and confirmed the working tree was left clean.

Evidence: Manual PR generated brief excerpt

Generated brief excerpt shows: --skip=pr,ci, branch push through no-mistakes, no PR/CI monitoring, and manual PR-ready status.

Generated brief: /var/folders/t0/dws_m6l52lj97dkt9f4hd_sh0000gn/T/no-mistakes-evidence/01KVTKFHBWBD1CHR7ZPTDJ1T17/manual-pr-home/data/ship-evidence-prci/brief.md

# Definition of done
The task is complete only when committed on your branch.
When you believe it is complete, append `done: {summary}` to the status file and stop.
Firstmate will then instruct you to run the no-mistakes validation pipeline with `--skip=pr,ci`.
During validation, fix auto-fix findings yourself; escalate ask-user findings per rule 6.
No-mistakes may push your branch, but does not open a PR or monitor CI in this workflow.
After no-mistakes finishes validation and branch push, append `done: validated with --skip=pr,ci; branch pushed and ready for captain/manual PR` and stop. You are finished.

Evidence: Full generated crewmate brief

You are a crewmate: an autonomous worker agent managed by firstmate. Work on your own; do not wait for a human.

# Task
{TASK}

# Setup
You are in a disposable git worktree of app, at a detached HEAD on a clean default branch.
1. First action: create your branch: `git checkout -b fm/ship-evidence-prci`
2. Run `no-mistakes doctor`; if it reports the repo is not initialized here, run `no-mistakes init`.

# Rules
1. Never push to the default branch. Never open or merge a PR. Push only your `fm/ship-evidence-prci` branch, and only as part of the no-mistakes validation pipeline.
2. Stay inside this worktree; modify nothing outside it.
3. Use gh-axi for GitHub operations and chrome-devtools-axi for browser operations.
4. Report status by appending one line:
   `echo "{state}: {one short line}" >> '/var/folders/t0/dws_m6l52lj97dkt9f4hd_sh0000gn/T/no-mistakes-evidence/01KVTKFHBWBD1CHR7ZPTDJ1T17/manual-pr-home/state/ship-evidence-prci.status'`
   States: working, needs-decision, blocked, done, failed.
   Each append wakes firstmate, so report sparingly: only phase changes a supervisor
   would act on (setup done, bug reproduced, fix implemented, validation passed) and the
   needs-decision/blocked/done/failed states. No step-by-step FYI progress lines;
   firstmate reads your pane for that.
5. If you hit the same obstacle twice, append `blocked: {why}` and stop; firstmate will help.
6. If a decision belongs to a human (product choices, destructive actions, ask-user findings),
   append `needs-decision: {summary of options}` and stop. Firstmate will reply with the decision.

# Project memory
If `AGENTS.md` or `CLAUDE.md` already exists, or if this task produced durable project-intrinsic knowledge, run `/Users/michael/.no-mistakes/worktrees/458d9fed4d70/01KVTKFHBWBD1CHR7ZPTDJ1T17/bin/fm-ensure-agents-md.sh .` in the worktree.
If this task produced durable project-intrinsic knowledge, record it in `AGENTS.md` as part of your change.
Keep it proportionate: skip `AGENTS.md` edits for trivial tasks that produced no durable project knowledge.

# Definition of done
The task is complete only when committed on your branch.
When you believe it is complete, append `done: {summary}` to the status file and stop.
Firstmate will then instruct you to run the no-mistakes validation pipeline with `--skip=pr,ci`.
During validation, fix auto-fix findings yourself; escalate ask-user findings per rule 6.
No-mistakes may push your branch, but does not open a PR or monitor CI in this workflow.
After no-mistakes finishes validation and branch push, append `done: validated with --skip=pr,ci; branch pushed and ready for captain/manual PR` and stop. You are finished.

Evidence: Evidence transcript

Generated by: FM_HOME=/var/folders/t0/dws_m6l52lj97dkt9f4hd_sh0000gn/T/no-mistakes-evidence/01KVTKFHBWBD1CHR7ZPTDJ1T17/manual-pr-home bin/fm-brief.sh ship-evidence-prci app
Generated brief: /var/folders/t0/dws_m6l52lj97dkt9f4hd_sh0000gn/T/no-mistakes-evidence/01KVTKFHBWBD1CHR7ZPTDJ1T17/manual-pr-home/data/ship-evidence-prci/brief.md

# Rules
1. Never push to the default branch. Never open or merge a PR. Push only your `fm/ship-evidence-prci` branch, and only as part of the no-mistakes validation pipeline.
2. Stay inside this worktree; modify nothing outside it.
3. Use gh-axi for GitHub operations and chrome-devtools-axi for browser operations.
4. Report status by appending one line:
   `echo "{state}: {one short line}" >> '/var/folders/t0/dws_m6l52lj97dkt9f4hd_sh0000gn/T/no-mistakes-evidence/01KVTKFHBWBD1CHR7ZPTDJ1T17/manual-pr-home/state/ship-evidence-prci.status'`
   States: working, needs-decision, blocked, done, failed.
   Each append wakes firstmate, so report sparingly: only phase changes a supervisor
   would act on (setup done, bug reproduced, fix implemented, validation passed) and the
   needs-decision/blocked/done/failed states. No step-by-step FYI progress lines;
   firstmate reads your pane for that.
5. If you hit the same obstacle twice, append `blocked: {why}` and stop; firstmate will help.
6. If a decision belongs to a human (product choices, destructive actions, ask-user findings),
   append `needs-decision: {summary of options}` and stop. Firstmate will reply with the decision.

# Project memory
If `AGENTS.md` or `CLAUDE.md` already exists, or if this task produced durable project-intrinsic knowledge, run `/Users/michael/.no-mistakes/worktrees/458d9fed4d70/01KVTKFHBWBD1CHR7ZPTDJ1T17/bin/fm-ensure-agents-md.sh .` in the worktree.
If this task produced durable project-intrinsic knowledge, record it in `AGENTS.md` as part of your change.
Keep it proportionate: skip `AGENTS.md` edits for trivial tasks that produced no durable project knowledge.

# Definition of done
The task is complete only when committed on your branch.
When you believe it is complete, append `done: {summary}` to the status file and stop.
Firstmate will then instruct you to run the no-mistakes validation pipeline with `--skip=pr,ci`.
During validation, fix auto-fix findings yourself; escalate ask-user findings per rule 6.
No-mistakes may push your branch, but does not open a PR or monitor CI in this workflow.
After no-mistakes finishes validation and branch push, append `done: validated with --skip=pr,ci; branch pushed and ready for captain/manual PR` and stop. You are finished.

- Outcome: 🔧 1 issue found → auto-fixed ✅ across 2 runs (13m40s)

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

🔧 **Review** - 2 issues found → auto-fixed ✅

• 🚨 AGENTS.md:421 - The new rule requires every no-mistakes instruction to include --skip=push,pr,ci, but generated no-mistakes briefs still say firstmate will run /no-mistakes to ship a PR and that the crewmate should report done: PR {url} checks green after CI. New tasks will get conflicting instructions and may still push/open PRs despite this workflow change; update bin/fm-brief.sh's no-mistakes definition of done to match the manual PR flow.
• ⚠️ bin/fm-bootstrap.sh:386 - Appending to a valid single-line TOML array with a trailing comma, such as post_create = ["existing",], produces post_create = ["existing",, "<hook>"], corrupting the Treehouse config. Strip an existing trailing comma before inserting, or rewrite the array through a safer parser path.

🔧 Fix: Align generated no-mistakes brief contract
✅ Re-checked - no issues remain.

🔧 **Test** - 1 issue found → auto-fixed ✅

• 🚨 AGENTS.md:421 - Captain, the implemented no-mistakes instructions skip the push step too: AGENTS.md requires --skip=push,pr,ci and says no-mistakes skips branch push. The generated crewmate brief mirrors that by telling the crewmate never to push and to stop before manual push+PR. That does not demonstrate the stated intent of validation running through push while only PR/CI are skipped with --skip=pr,ci.
• git diff 2a2fe0566e63b883ec53dc3e34dc14306689f989..0b9ac82dd25e61de49e59221918dce3dbab2f965 -- AGENTS.md bin/fm-brief.sh tests/fm-brief.test.sh README.md .github/workflows/no-mistakes-required.yml
• tests/fm-brief.test.sh
• FM_HOME=/var/folders/t0/dws_m6l52lj97dkt9f4hd_sh0000gn/T/no-mistakes-evidence/01KVTKFHBWBD1CHR7ZPTDJ1T17/home bin/fm-brief.sh ship-test app
• nl -ba AGENTS.md | sed -n '406,440p' and nl -ba bin/fm-brief.sh | sed -n '184,198p'
• git status --short

🔧 Fix: Captain, allow no-mistakes push before manual PR
✅ Re-checked - no issues remain.

• sed -n '1,220p' /Users/michael/.agents/skills/no-mistakes/SKILL.md
• pwd; git status --short; git rev-parse --show-toplevel; git log --oneline --decorate -5
• git diff --stat 2a2fe0566e63b883ec53dc3e34dc14306689f989..HEAD; git diff --name-only 2a2fe0566e63b883ec53dc3e34dc14306689f989..HEAD
• sed -n '380,445p' AGENTS.md; sed -n '170,205p' bin/fm-brief.sh; sed -n '1,260p' tests/fm-brief.test.sh
• tests/fm-brief.test.sh && tests/fm-bootstrap.test.sh && tests/fm-spawn-treehouse-env.test.sh && tests/fm-treehouse-post-create.test.sh
• for t in tests/*.test.sh; do echo "== $t =="; "$t" || exit $?; done
• FM_HOME=/var/folders/t0/dws_m6l52lj97dkt9f4hd_sh0000gn/T/no-mistakes-evidence/01KVTKFHBWBD1CHR7ZPTDJ1T17/manual-pr-home bin/fm-brief.sh ship-evidence-prci app
• grep checks on the generated brief for --skip=pr,ci, branch-push wording, manual PR-ready status, and absence of --skip=push,pr,ci / old PR+CI status wording
• git status --short && find . -maxdepth 1 -type d -name '.tmp.*' -print

⏭️ **Document** - skipped

• ⚠️ .github/workflows/no-mistakes-required.yml:18 - Workflow UI/error text still says PRs must be raised through git push no-mistakes and that the pipeline writes the PR body. I left it unchanged because it is executable workflow configuration, not a documentation file or doc comment.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.