fix(bin): raise axi-family bootstrap floors to current-latest

[marcusvpn1]

Intent

The captain decided this fork should diverge from the upstream kunchenguid/firstmate template rather than track it closely, but wants the ~10 low-risk, isolated fixes identified in the diff review (data/fm-upstream-firstmate-diff-review/report.md) cherry-picked now. Each fix must be verified as genuinely isolated - not touching fork-unique work (the AGY headless-crewmate harness adapter, codebase-memory MCP integration via Pi extension, progress-ledger skill, and spec-scaffold skill) - before applying. If a candidate turns out not to be as isolated as the review assessed, skip it and note why rather than forcing it in.

Assessment against the current fork tip found 7 of the 10 candidates already present (514f0ab Bash 3.2 brief scaffolding, 41ffd45 CLAUDE_CONFIG_DIR forwarding, e595611 .gitignore config/ directory, a117b41 Claude session-lock ancestry, c0c0881 Calm version ceiling, 7cbb3f6 subagent todo tools, 4a9979a stale AXI tools - the last already landed as fork commit 763179a), 1 remote-only (bea3d23 fm-remote-entrypoint.sh symlink fix, whose file is absent in this fork which has no remote secondmates), and 2 genuinely missing. Those 2 were applied: ef2c3a2 raises the axi-family bootstrap floors to current-latest (lavish-axi 0.1.45, tasks-axi 0.2.4, quota-axi 0.1.17, gh-axi 0.1.29 unchanged) and states the floor policy beside the constants, and 70aeba8 raises lavish-axi to 0.1.46. Conflict resolution dropped the remote-secondmate hunks (bin/fm-backlog-receive.sh, tests/fm-on.test.sh, tests/fm-remote-doctor.test.sh) that do not exist in this fork.

What Changed

• Raised the bootstrap version floors for the axi-family tools to their current-latest releases — lavish-axi 0.1.46, tasks-axi 0.2.4, and quota-axi 0.1.17 — and added the previously missing lavish-axi floor check so below-floor builds report MISSING.
• Centralized the floor policy beside the version constants in bin/fm-bootstrap.sh; the per-tool libs (fm-tasks-axi-lib.sh, fm-quota-axi-lib.sh), bootstrap diagnostics, and docs now reference that policy instead of restating per-tool rationale.
• Updated test fixtures to version-aware stubs via a new fm_fake_version_tool helper, pinned the new floors across fixtures, and added a readiness-file handshake in the watcher-lock test to remove a startup race.

Risk Assessment

✅ Low: The change is well-bounded and mechanical: it raises three axi-family bootstrap floors and adds a lavish-axi floor gate with a documented policy, updates docs and test fixtures consistently, and includes one correct test-flakiness fix, with no defects, stale references, or intent contradictions found.

Testing

I ran the bootstrap behavior suite and a manual end-to-end exercise of bin/fm-bootstrap.sh against a fake toolchain, confirming each raised axi-family floor (lavish-axi 0.1.46, tasks-axi 0.2.4, quota-axi 0.1.17, gh-axi unchanged at 0.1.29) reports the exact MISSING install line below the floor and is silent at/above it, and that the newly-added lavish-axi gate flips an old 0.1.45 build from silently accepted (base) to MISSING (target); all touched test suites pass, with only a pre-existing watcher-lock timing test flaking once before passing on retry.

Evidence: axi-floor bootstrap transcript (MISSING vs silent at each boundary)

=== lavish-axi floor (LAVISH_AXI_MIN=0.1.46) === lavish-axi 0.1.45 (below floor) => MISSING: lavish-axi (install: npm install -g lavish-axi && lavish-axi setup hooks) lavish-axi 0.1.46 (at floor) => (silent - accepted) lavish-axi 0.1.47 (above floor) => (silent - accepted) === tasks-axi floor (FM_TASKS_AXI_MIN=0.2.4) === tasks-axi 0.2.3 (below floor) => MISSING: tasks-axi (install: npm install -g tasks-axi) tasks-axi 0.2.4 (at floor) => (silent - accepted) === quota-axi floor (FM_QUOTA_AXI_MIN=0.1.17) === quota-axi 0.1.16 (below floor) => MISSING: quota-axi (install: npm install -g quota-axi) quota-axi 0.1.17 (at floor) => (silent - accepted) === gh-axi floor unchanged (GH_AXI_MIN=0.1.29) === gh-axi 0.1.29 (at unchanged floor) => (silent - accepted)

=== lavish-axi floor (LAVISH_AXI_MIN=0.1.46) ===
lavish-axi 0.1.45 (below floor)                      => MISSING: lavish-axi (install: npm install -g lavish-axi && lavish-axi setup hooks)
lavish-axi 0.1.46 (at floor)                         => (silent - accepted)
lavish-axi 0.1.47 (above floor)                      => (silent - accepted)

=== tasks-axi floor (FM_TASKS_AXI_MIN=0.2.4) ===
tasks-axi 0.2.3 (below floor)                        => MISSING: tasks-axi (install: npm install -g tasks-axi)
tasks-axi 0.2.4 (at floor)                           => (silent - accepted)

=== quota-axi floor (FM_QUOTA_AXI_MIN=0.1.17) ===
quota-axi 0.1.16 (below floor)                       => MISSING: quota-axi (install: npm install -g quota-axi)
quota-axi 0.1.17 (at floor)                          => (silent - accepted)

=== gh-axi floor unchanged (GH_AXI_MIN=0.1.29) ===
gh-axi 0.1.29 (at unchanged floor)                   => (silent - accepted)

Evidence: lavish-axi floor before/after (base silently accepts 0.1.45, target reports MISSING)

base (lavish-axi 0.1.45) => (silent - accepted) target (lavish-axi 0.1.45) => MISSING: lavish-axi (install: npm install -g lavish-axi && lavish-axi setup hooks)

base (lavish-axi 0.1.45)     => (silent - accepted)
target (lavish-axi 0.1.45)   => MISSING: lavish-axi (install: npm install -g lavish-axi && lavish-axi setup hooks)

- Outcome: ⚠️ 1 info across 1 run (48m48s)

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

✅ **Review** - passed

✅ No issues found.

⚠️ **Test** - 1 info

• ℹ️ tests/fm-watcher-lock.test.sh:334 - test_lock_empty_pid_uses_minimum_grace failed once on the first run ('rc=0 held=' - the freshly-created empty lock dir was stolen with zero stale threshold), then passed on three consecutive reruns. The test and bin/fm-wake-lib.sh lock logic are byte-identical between base and target commits (this diff only touched test_watch_restart_attaches_to_healthy_peer's peer-readiness signaling), so this is a pre-existing timing flake under load, not a regression introduced by the axi-floor change.
• bash tests/fm-bootstrap.test.sh
• bash tests/fm-session-start.test.sh
• bash tests/fm-teardown.test.sh
• bash tests/fm-x-mode.test.sh
• bash tests/fm-startup-memory-budget.test.sh
• bash tests/fm-shared-captain-inheritance.test.sh
• bash tests/fm-secondmate-harness.test.sh
• bash tests/fm-secondmate-liveness.test.sh
• bash tests/fm-secondmate-sync.test.sh
• bash tests/fm-watcher-lock.test.sh (flaked once, passed on 3 reruns)
• bash /tmp/fm-axi-floor-demo.sh (runs real bin/fm-bootstrap.sh with fake lavish-axi/tasks-axi/quota-axi/gh-axi at boundary versions)
• FM_BASE_ROOT=<base-commit-tree> bash /tmp/fm-axi-before-after.sh (real base vs target fm-bootstrap.sh against lavish-axi 0.1.45)

✅ **Document** - passed

✅ No issues found.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.