Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Declare a structural intent BEFORE editing in a fleet — what symbols you'll touch and why — so other agents on your branch coordinate around you. Use when the user says 'I'm about to edit/rename/refactor X', or when starting any non-trivial edit while other agents share your repo+branch. Returns the graph blast radius, overlapping live intents on your branch, and a shift-left coordination/partition hint. Do not start editing shared symbols without publishing first.
.claude/skills/syncable-dev-memtrace-fleet-publish-intent/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 9% | 0% |
| case-16 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-20 | ✗→✓ | ▲ Improved | -17% | 0% |
fleet_publish_intent is step 1 of the fleet protocol: announce what you're about to touch so other agents on the same (repo, branch) coordinate around you. It's a typed, ~20-token declaration — not prose.
jsoncfleet_publish_intent({ repo_id: "myrepo", branch: "session/auth-revamp", // your fleet's session branch agent_id:"agent-a", touched: ["auth::verify_token"], // qualified symbol identities intent: {"refactor": {"pattern": "change_signature"}}, assignment: "widen verify_token signature for pagination" // the alignment anchor })
Full parameter spec for every Memtrace tool: references/mcp-parameters.md (bundled at the memtrace-skills plugin root).
json{ "impact_preview": "…graph blast radius of the touched symbols…", "active_conflicts": [], "coordination": { "would_escalate": false, "suggested_partition": null } }
impact_preview — the real graph blast radius of the touched symbols.active_conflicts — overlapping live intents on your branch (none from otherbranches; coordination is branch-scoped).
coordination — a shift-left hint: would_escalate, a suggested_partition(who should own a contested symbol), and advice to realign before you edit.
{"refactor":{"pattern":"rename_symbol"|"change_signature"|"move_symbol"|"extract_function"|…}}{"feature_add":{"surface":"new_symbol"|"new_endpoint"|…}}{"bug_fix":{"defect":"logic_error"|"null_handling"|…}}{"cleanup":{"kind":"dead_code"|"unused_import"|…}}{"performance":{"axis":"latency"|…}}, {"security_fix":{"severity":"high"}},{"test_add":{"covers":[…]}}, "docs_only", "exploratory"
Destructive kinds — change_signature, move_symbol, cleanup/dead_code — are what raise a Class C decision when they overlap another agent's work.
branch (your session branch) and assignment (your task).fleet_preflight (same inputs, no registration).active_conflicts is good — proceed and fleet_record_episode when done.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 12,552 | 6,699 | -47% | 1 | 1 | 0% | 1,007 | 1,177 | +17% | 0 | 0 | — |
case-02 | fail→fail | 13,824 | 3,834 | -72% | 1 | 1 | 0% | 1,869 | 1,216 | -35% | 0 | 0 | — |
case-03 | fail→fail | 6,399 | 6,885 | +8% | 1 | 1 | 0% | 449 | 1,107 | +147% | 0 | 0 | — |
case-04 | fail→fail | 13,982 | 5,275 | -62% | 1 | 1 | 0% | 934 | 940 | +1% | 0 | 0 | — |
case-05 | fail→fail | 4,282 | 6,875 | +61% | 1 | 1 | 0% | 619 | 1,120 | +81% | 0 | 0 | — |
case-06 | fail→fail | 4,724 | 5,479 | +16% | 1 | 1 | 0% | 621 | 1,004 | +62% | 0 | 0 | — |
case-17 | fail→pass | 8,430 | 1,772 | -79% | 1 | 1 | 0% | 1,254 | 982 | -22% | 0 | 0 | — |
case-07 | fail→fail | 6,572 | 5,656 | -14% | 1 | 1 | 0% | 1,145 | 953 | -17% | 0 | 0 | — |
case-08 | fail→fail | 4,888 | 5,893 | +21% | 1 | 1 | 0% | 832 | 1,121 | +35% | 0 | 0 | — |
case-09 | fail→fail | 8,047 | 6,487 | -19% | 1 | 1 | 0% | 665 | 1,104 | +66% | 0 | 0 | — |
case-10 | pass→pass | 6,290 | 4,955 | -21% | 1 | 1 | 0% | 1,176 | 995 | -15% | 0 | 0 | — |
case-15 | fail→fail | 5,598 | 6,103 | +9% | 1 | 1 | 0% | 874 | 990 | +13% | 0 | 0 | — |
case-11 | fail→pass | 5,640 | 3,218 | -43% | 1 | 1 | 0% | 881 | 963 | +9% | 0 | 0 | — |
case-12 | fail→fail | 4,941 | 5,457 | +10% | 1 | 1 | 0% | 884 | 930 | +5% | 0 | 0 | — |
case-13 | fail→fail | 2,710 | 5,876 | +117% | 1 | 1 | 0% | 466 | 969 | +108% | 0 | 0 | — |
case-14 | pass→fail | 7,149 | 5,600 | -22% | 1 | 1 | 0% | 1,267 | 952 | -25% | 0 | 0 | — |
case-16 | fail→pass | 10,891 | 3,066 | -72% | 1 | 1 | 0% | 1,904 | 1,142 | -40% | 0 | 0 | — |
case-18 | pass→pass | 11,017 | 2,745 | -75% | 1 | 1 | 0% | 1,913 | 880 | -54% | 0 | 0 | — |
case-19 | fail→pass | 6,903 | 1,764 | -74% | 1 | 1 | 0% | 1,028 | 939 | -9% | 0 | 0 | — |
case-20 | fail→pass | 8,122 | 3,265 | -60% | 1 | 1 | 0% | 1,317 | 1,094 | -17% | 0 | 0 | — |
case-21 | fail→fail | 5,499 | 5,749 | +5% | 1 | 1 | 0% | 984 | 1,012 | +3% | 0 | 0 | — |
case-22 | fail→pass | 12,343 | 3,391 | -73% | 1 | 1 | 0% | 2,292 | 1,310 | -43% | 0 | 0 | — |
case-23 | fail→pass | 4,646 | 3,190 | -31% | 1 | 1 | 0% | 736 | 1,221 | +66% | 0 | 0 | — |
case-24 | fail→pass | 5,905 | 2,706 | -54% | 1 | 1 | 0% | 1,001 | 1,097 | +10% | 0 | 0 | — |
case-25 | pass→pass | 4,202 | 4,752 | +13% | 1 | 1 | 0% | 719 | 1,426 | +98% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted, and 12 counted toward the lift figure. The other 13 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +28 percentage points is the difference between those two pass rates over the 12 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.