feat(supervision): add injection/honeypot guard for external-repo crewmate PRs

[e-jung]

What Changed

• Add a review-stage injection/honeypot guard (bin/fm-injection-scan.sh) that scans a crewmate's added diff lines for planted prompt-injection payloads (notice/marker files, self-incriminating reveal text, hidden HTML-comment or zero-width instructions, base64 blobs, "ignore previous instructions") before firstmate relays an external-repo PR upstream, paired with a --fork-pr brief scaffold that treats the target repo's agent/contributor files as untrusted data.
• Harden fleet supervision: fm-spawn.sh now refuses crewmate/scout launches without an explicit harness when config/crew-dispatch.json is active (and bootstrap surfaces the active rules), and the new bin/fm-config-push.sh pushes inheritable local config into live secondmate homes without a full bootstrap.
• Support attaching a single image to X replies and completion follow-ups through fm-x-reply.sh and the new fm-x-lib.sh (media-type detected, base64-encoded, bound to the opener tweet).

Risk Assessment

⚠️ Medium: Substantial, security-sensitive new logic (a deterministic injection/honeypot scanner for external-repo PRs, public X image replies with base64 encoding, a dispatch-profile enforcement backstop, and a new config-push script) is implemented defensively and covered by comprehensive hermetic tests with no blocking correctness or security bugs found; safe to merge, with the security-relevant paths benefiting from a human eyeball during the normal gate.

Testing

Completed 1 recorded test check.

• Outcome: 🔧 1 issue found → auto-fixed ✅ across 2 runs (15m52s)

Pipeline

Updates from git push no-mistakes

⏭️ **intent** - skipped

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

⚠️ **Review** - 2 infos

• ℹ️ bin/fm-x-lib.sh:231 - fmx_reply_outbox_json has 8 near-identical jq branches (followup x image x single/thread) and fmx_reply_payload_json has 4, differing only in which keys are emitted. I verified all 8 outbox branches produce correct output, but if the relay payload contract changes every branch must be hand-updated consistently. These could be collapsed by building a base object and conditionally merging keys, e.g. .[0] as $t | {request_id:$rid,text:($t//"")} | if $n>1 then .texts=$chunks else . end | if $followup=1 then .endpoint="followup" else . end | if $image!="" then .image=($image|fromjson?) else . end. Not a correctness issue.
• ℹ️ bin/fm-config-push.sh:101 - The read loop binds the 4th pipe field as meta (without an underscore prefix, unlike _window) but never references it in the loop body - only id and home are used. Cosmetic inconsistency; the live_secondmate_meta_records helper emits the meta-file path as its 4th field for callers that need it, but this caller resolves homes via validate_secondmate_home instead.

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

• 🚨 tests failed with exit code 1
• command -v tmux >/dev/null || { echo "tmux is required for e2e tests" >&2; exit 1; }; tmux -V; rc=0; for t in tests/*.test.sh; do echo "== $t =="; bash "$t" || rc=1; done; exit "$rc"

🔧 Fix: Add teardown merge-tree fallback for old git
✅ Re-checked - no issues remain.

• command -v tmux >/dev/null || { echo "tmux is required for e2e tests" >&2; exit 1; }; tmux -V; rc=0; for t in tests/*.test.sh; do echo "== $t =="; bash "$t" || rc=1; done; exit "$rc"

✅ **Document** - passed

✅ No issues found.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.

[kunchenguid]

thanks for this. two things before it can land:

1. it needs to come through no-mistakes - that's why the "PR must be raised via no-mistakes" check is red (this was opened directly). the reason i ask: no-mistakes runs review, tests, lint, and docs on your branch before the PR, so contributions arrive already validated - it keeps the quality bar high and saves me hand-reviewing every change, which matters since i maintain this solo. re-raise by running the pipeline on your branch and it'll update the PR through the gate.
2. CI is also showing real test failures (the OS matrix + e2e), so those need a fix as part of the re-raise.

the change looks useful, it just needs to come through that path green. ping me if the setup trips you up.

[kunchenguid]

Thanks for the PR! It looks like this branch has a merge conflict with the base branch right now. When you get a chance, could you rebase onto (or merge in) the latest base branch, resolve the conflict, and push? Once GitHub shows the PR as mergeable again, it'll be picked back up for review.

_{Noted for firstmate#67 at b7c282af.}

[kunchenguid]

Quick reminder: this PR still looks blocked on a rebase or merge conflict fix.

If you are still interested, please rebase onto the current base branch, resolve the conflict, and push.

If I do not hear back, I may close this as inactive.

[kunchenguid]

I am closing this because it has been waiting on a rebase or merge-conflict fix since 2026-07-01, and I have not seen a comment or push since then.

If you still want to keep working on this, please reopen it or open a new PR and mention this one.

Happy to take another look when there is an update.