servo

Run a headless agent loop against a scaffolded target. Subprocesses `claude -p --agent <runner|judge>` against the target N iterations under hard guardrails (iteration cap, cumulative cost ceiling, context-fill refusal, oracle-score plateau detection, SIGINT/SIGTERM cleanup), uses `gate.py` as the truth-source after each iteration, and checkpoints state to `<target>/.servo/runs/<run-id>/state.json` so an interrupted or capped run can be resumed. Fire this skill when the user wants to: - "run an agent loop" / "iterate on this codebase" / "headless loop" - "let claude iterate against my target" / "fire-and-forget claude on this" - "resume the loop" / "continue the prior loop" / "pick up where it left off" - "loop with cost ceiling X" / "loop until the oracle passes" - "use /goal to keep going" / "goal-driven loop" / "--driver goal" — the ADR-0008 mode that delegates continuation to Claude Code's `/goal` while servo keeps the oracle + guardrail layer (caps on the outer call, final `gate.p

Run a headless agent loop against a scaffolded target. Subprocesses `claude -p --agent <runner|judge>` against the target N iterations under hard guardrails (iteration cap, cumulative cost ceiling, context-fill refusal, oracle-score plateau detection, SIGINT/SIGTERM cleanup), uses `gate.py` as the truth-source after each iteration, and checkpoints state to `<target>/.servo/runs/<run-id>/state.json` so an interrupted or capped run can be resumed.

Run a headless agent loop against a scaffolded target. Subprocesses `claude -p --agent <runner|judge>` against the target N iterations under hard guardrails (iteration cap, cumulative cost ceiling, context-fill refusal, oracle-score plateau detection, SIGINT/SIGTERM cleanup), uses `gate.py` as the truth-source after each iteration, and checkpoints state to `<target>/.servo/runs/<run-id>/state.json` so an interrupted or capped run can be resumed. Fire this skill when the user wants to: - "run an agent loop" / "iterate on this codebase" / "headless loop" - "let claude iterate against my target" / "fire-and-forget claude on this" - "resume the loop" / "continue the prior loop" / "pick up where it left off" - "loop with cost ceiling X" / "loop until the oracle passes" - "use /goal to keep going" / "goal-driven loop" / "--driver goal" — the ADR-0008 mode that delegates continuation to Claude Code's `/goal` while servo keeps the oracle + guardrail layer (caps on the outer call, final `gate.p

Author a frozen text content-fidelity eval component for a project — the non-visual sibling of servo's design-eval. Use when generated text (a reply, a doc, a templated message) must match an intended rubric or spec ("does this text match the intended tone/content/style?") and that fidelity should drive or gate the loop. Gathers the generated text under test (file read or generator command), judges fidelity with a pinned text-capable model (n-sampled, lower-bound), freezes the definition, and installs a `score_content_fidelity` component into the project's `oracle.sh`. Do not use for deterministic checks (use scaffold-init / spec-oracle), for visual/UI fidelity (use design-eval), or for the per-iteration judge agent.

Author a frozen UI design-fidelity eval component for a project — the non-deterministic sibling of servo's deterministic component templates. Use when a project must be built to match a design mockup ("does the UI match the design?") and that fidelity should drive or gate the loop. Captures app vs reference screenshots, judges fidelity with a pinned vision model (n-sampled, lower-bound), freezes the definition, and installs a `score_design_fidelity` component into the project's `oracle.sh`. Do not use for deterministic checks (use scaffold-init / spec-oracle) or for the per-iteration judge agent.

Decide whether an engineering spec is suitable for Evaluation-Driven Development before any unattended loop runs against it. Emits the ADR-0015 **suitability verdict** — a closed three-state gate (`suitable` / `needs_evidence` / `unsuitable`), fail-closed, with an actionable `missing_evidence` checklist — written to `<target>/.servo/suitability/<spec-id>.json`. A gate, not a score: the loop never optimizes toward a meaningless green oracle on work that cannot be evaluated. Fire this skill when the user wants to: - "is this spec/work suitable for EDD?" / "should this go through servo?" - "what evidence is this spec missing?" / "why isn't this ready to compile?" - "why won't Compile proceed on this spec?" / "explain the suitability verdict" - "re-check suitability after I added tests" / "is <spec> EDD-shaped yet?" Do NOT fire on: - "scaffold the oracle" / "install servo" / "detect signals" — that's `/servo:scaffold-init` (001). Suitability *reads* the signals it detects; it does not synthe

Author a frozen text-judged eval from a spec-006 `residual_judgment` AC or a free-form goal. Guides triage, rubric shaping, reference-set collection, frozen n/δ/threshold/judge parameters, ADR-0024 freeze/install, and advisory judge audit. Use to author an eval, turn a goal into acceptance criteria, shape a rubric or reference set, freeze/install an eval, or audit judge trust. Do NOT use to run evals (`quality-gate`/`agent-loop`), classify deterministic checks (`spec-oracle`), decide EDD suitability (`edd-suitability`), or build design/content fidelity evals (use the `design-eval` / `content-fidelity` presets). This is the generic surface for other text-judged criteria.

Assemble the [ADR-0016](../../docs/decisions/adr-0016-execution-plan-artifact.md) **execution plan** — the durable Compile→Run handoff artifact — from the Compile inputs already produced upstream (the suitability verdict, the oracle, the spec-oracle overlay) and write it to `<target>/.servo/plans/<spec-id>/plan.json`. This is the **last step of Servo Compile**: it decides nothing new, classifies nothing, and adds no gate — it only references what Compile already produced. Fire this skill when the user wants to: - "compile the execution plan for this spec" - "produce the Compile→Run handoff" - "emit `plan.json` for this spec" - "assemble the ADR-0016 plan for <spec>" / "wire up the Compile→Run artifact" Do NOT fire on: - "is this spec suitable for EDD?" / "what evidence is missing?" — that's `/servo:edd-suitability` (015). This skill *consumes* the suitability verdict as a hard precondition; it does not decide it. - "classify the acceptance criteria" / "build the spec-oracle overlay" —

Run heartbeat or set up a scheduled heartbeat for a servo-scaffolded target. The heartbeat is servo's Routine-triggered front-end: discover project signals into a triage inbox, read the heartbeat inbox, dispatch heartbeat findings, or run heartbeat end to end under one whole-heartbeat cost ceiling. Fire this skill for requests mentioning a Routine, cron, GitHub Actions schedule, or a scheduled agent that should invoke servo on an interval. Do NOT fire on: - "run the oracle" / "score this code" / "oracle score" — that's `/servo:quality-gate`. - "iterate on this" / "run an agent loop" / "fix this manually with the loop" — that's `/servo:agent-loop`. - "set up servo" / "scaffold the oracle" / "install servo in this project" — that's `/servo:scaffold-init`. - "install the oracle hook" / "oracle hook status" — that's `/servo:oracle-hook`. - "spec oracle" / "turn this spec into checks" — that's `/servo:spec-oracle`. - "design eval" / "visual fidelity score" — that's `/servo:design-

Install, uninstall, or report the status of servo's meta-judge `Stop` hook in a target project. The installed hook scores every assistant turn against the scaffolded oracle (via `gate.py`) and blocks the stop with a structured retry hint when the oracle is below threshold — the interactive cousin of `/servo:agent-loop`. Installing it mutates `<target>/.claude/settings.json`, so it is a **Tier-2** surface: offered explicitly, never auto-installed. Fire this skill when the user wants to: - "install the oracle hook" / "add the Stop hook" / "set up the meta-judge" - "uninstall the oracle hook" / "remove the meta-judge" / "remove the Stop hook" - "is the oracle hook installed?" / "oracle hook status" / "check the hook" Do NOT fire on: - "run the oracle" / "score this code" / "what's the oracle score?" — that's `/servo:quality-gate` (a one-shot score, not the hook installer). - "iterate on this" / "run an agent loop" / "headless loop" — that's `/servo:agent-loop` (the headless cousin; this hoo

Run the scaffolded `oracle.sh` against a target project and surface a closed-contract result (0 = pass / 1 = below threshold / 2 = env error). Also exposes an `audit` subcommand that prints the servo install manifest without invoking the oracle. Fire this skill when the user wants to: - "score this code" / "score this project" / "run quality gate" - "run the oracle" / "what's the oracle score?" - "check the oracle" / "did the oracle pass?" - "show me the gate result" / "is the gate passing?" - "what does this servo install include?" (→ audit subcommand) Do NOT fire on: - "set up servo" / "scaffold oracle" / "install servo" — those are `/servo:scaffold-init`'s territory (a sibling skill, not this one). - "fix the failing test" / "make tests pass" / "debug this test" — out of scope. The gate reports; it does not modify code or tests. - "run my tests" / "lint this" — call the underlying tool directly; the gate is the composite scorer, not a single-tool runner. - "review my code"

Compile a target project's signals into an executable oracle — the first Servo Compile step of servo's Evaluation-Driven Development engine. Probes the target for signals (tests, lint, CI, language) and drops a tailored `oracle.sh`, `.servo/install.json` manifest, and `.servo/refinement-todo.md` deferred-decisions list. Fire this skill when the user wants to: - "set up servo" or "set up servo on this project" - "scaffold oracle for this project" or "scaffold an oracle" - "install agent loop infrastructure" - "install servo here" - "bring this repo under servo's closed-loop control" Do NOT fire on: - bare "oracle" or "score this code" — those refer to *running* an already-scaffolded oracle.sh, which is a future runtime skill (`/servo:quality-gate`), not this scaffolder. - "review my code" / "run my tests" / "lint this" — out of scope. - "set up jig" / "scaffold jig" — that's jig's `/jig:scaffold-init`, a sibling plugin, not this one. When in doubt, ask the user which servo skil

Compile a spec or slice into a reviewable, project-owned **evidence overlay**: map each acceptance criterion to a deterministic check family, write a `plan.md` + `checks.json` under the spec's own `oracle/<id>/` directory, compile the checks into an installable `oracle.sh` component, and freeze it behind an explicit approval so an unattended loop can optimize against the spec's promises instead of a hand-waved single judge. Fire this skill when the user wants to: - "generate an oracle for this spec" / "spec oracle for <spec>" - "turn this spec/slice into checks" / "compile acceptance criteria into checks" - "evaluate / specify an oracle for a spec" / "what would servo judge for this spec?" - "plan the evidence for <spec>" / "install the spec overlay" / "approve the spec oracle" Do NOT fire on: - "run the oracle" / "score this code" / "what's the oracle score?" — that's `/servo:quality-gate`. This skill *builds and installs* the overlay; quality-gate *runs* it. - "iterate on this" / "ru

Author a frozen text content-fidelity eval component for a project — the non-visual sibling of servo's design-eval. Use when generated text (a reply, a doc, a templated message) must match an intended rubric or spec ("does this text match the intended tone/content/style?") and that fidelity should drive or gate the loop. Gathers the generated text under test (file read or generator command), judges fidelity with a pinned text-capable model (n-sampled, lower-bound), freezes the definition, and installs a `score_content_fidelity` component into the project's `oracle.sh`. Do not use for deterministic checks (use scaffold-init / spec-oracle), for visual/UI fidelity (use design-eval), or for the per-iteration judge agent.

Author a frozen UI design-fidelity eval component for a project — the non-deterministic sibling of servo's deterministic component templates. Use when a project must be built to match a design mockup ("does the UI match the design?") and that fidelity should drive or gate the loop. Captures app vs reference screenshots, judges fidelity with a pinned vision model (n-sampled, lower-bound), freezes the definition, and installs a `score_design_fidelity` component into the project's `oracle.sh`. Do not use for deterministic checks (use scaffold-init / spec-oracle) or for the per-iteration judge agent.

Decide whether an engineering spec is suitable for Evaluation-Driven Development before any unattended loop runs against it. Emits the ADR-0015 **suitability verdict** — a closed three-state gate (`suitable` / `needs_evidence` / `unsuitable`), fail-closed, with an actionable `missing_evidence` checklist — written to `<target>/.servo/suitability/<spec-id>.json`. A gate, not a score: the loop never optimizes toward a meaningless green oracle on work that cannot be evaluated.

Author a frozen text-judged eval from a spec-006 `residual_judgment` AC or a free-form goal. Guides triage, rubric shaping, reference-set collection, frozen n/δ/threshold/judge parameters, ADR-0024 freeze/install, and advisory judge audit. Use to author an eval, turn a goal into acceptance criteria, shape a rubric or reference set, freeze/install an eval, or audit judge trust. Do NOT use to run evals (`quality-gate`/`agent-loop`), classify deterministic checks (`spec-oracle`), decide EDD suitability (`edd-suitability`), or build design/content fidelity evals (use the `design-eval` / `content-fidelity` presets). This is the generic surface for other text-judged criteria.

Assemble the [ADR-0016](../../docs/decisions/adr-0016-execution-plan-artifact.md) **execution plan** — the durable Compile→Run handoff artifact — from the Compile inputs already produced upstream (the suitability verdict, the oracle, the spec-oracle overlay) and write it to `<target>/.servo/plans/<spec-id>/plan.json`. This is the **last step of Servo Compile**: it decides nothing new, classifies nothing, and adds no gate — it only references what Compile already produced.

Run heartbeat or set up a scheduled heartbeat for a servo-scaffolded target. The heartbeat is servo's Routine-triggered front-end: discover project signals into a triage inbox, read the heartbeat inbox, dispatch heartbeat findings, or run heartbeat end to end under one whole-heartbeat cost ceiling. Fire this skill for requests mentioning a Routine, cron, GitHub Actions schedule, or a scheduled agent that should invoke servo on an interval.

Install, uninstall, or report the status of servo's meta-judge `Stop` hook in a target project. The installed hook scores every assistant turn against the scaffolded oracle (via `gate.py`) and blocks the stop with a structured retry hint when the oracle is below threshold — the interactive cousin of `/servo:agent-loop`. Installing it mutates `<target>/.claude/settings.json`, so it is a **Tier-2** surface: offered explicitly, never auto-installed.

Run the scaffolded `oracle.sh` against a target project and surface a closed-contract result (0 = pass / 1 = below threshold / 2 = env error). Also exposes an `audit` subcommand that prints the servo install manifest without invoking the oracle.

Compile a target project's signals into an executable oracle — the first Servo Compile step of servo's Evaluation-Driven Development engine. Probes the target for signals (tests, lint, CI, language) and drops a tailored `oracle.sh`, `.servo/install.json` manifest, and `.servo/refinement-todo.md` deferred-decisions list.

Compile a spec or slice into a reviewable, project-owned **evidence overlay**: map each acceptance criterion to a deterministic check family, write a `plan.md` + `checks.json` under the spec's own `oracle/<id>/` directory, compile the checks into an installable `oracle.sh` component, and freeze it behind an explicit approval so an unattended loop can optimize against the spec's promises instead of a hand-waved single judge.

Author a frozen text content-fidelity eval component for a project — the non-visual sibling of servo's design-eval. Use when generated text (a reply, a doc, a templated message) must match an intended rubric or spec ("does this text match the intended tone/content/style?") and that fidelity should drive or gate the loop. Gathers the generated text under test (file read or generator command), judges fidelity with a pinned text-capable model (n-sampled, lower-bound), freezes the definition, and installs a `score_content_fidelity` component into the project's `oracle.sh`. Do not use for deterministic checks (use scaffold-init / spec-oracle), for visual/UI fidelity (use design-eval), or for the per-iteration judge agent.

Author a frozen text-judged eval from a spec-006 `residual_judgment` AC or a free-form goal. Guides triage, rubric shaping, reference-set collection, frozen n/δ/threshold/judge parameters, ADR-0024 freeze/install, and advisory judge audit. Use to author an eval, turn a goal into acceptance criteria, shape a rubric or reference set, freeze/install an eval, or audit judge trust. Do NOT use to run evals (`quality-gate`/`agent-loop`), classify deterministic checks (`spec-oracle`), decide EDD suitability (`edd-suitability`), or build design/content fidelity evals (use the `design-eval` / `content-fidelity` presets). This is the generic surface for other text-judged criteria.

Compile a target project's signals into an executable oracle — the first Servo Compile step of servo's Evaluation-Driven Development engine. Probes the target for signals (tests, lint, CI, language) and drops a tailored `oracle.sh`, `.servo/install.json` manifest, and `.servo/refinement-todo.md` deferred-decisions list. Fire this skill when the user wants to: - "set up servo" or "set up servo on this project" - "scaffold oracle for this project" or "scaffold an oracle" - "install agent loop infrastructure" - "install servo here" - "bring this repo under servo's closed-loop control" Do NOT fire on: - bare "oracle" or "score this code" — those refer to *running* an already-scaffolded oracle.sh, which is a future runtime skill (`/servo:quality-gate`), not this scaffolder. - "review my code" / "run my tests" / "lint this" — out of scope. - "set up jig" / "scaffold jig" — that's jig's `/jig:scaffold-init`, a sibling plugin, not this one. When in doubt, ask the user which servo skil

Assemble the [ADR-0016](../../docs/decisions/adr-0016-execution-plan-artifact.md) **execution plan** — the durable Compile→Run handoff artifact — from the Compile inputs already produced upstream (the suitability verdict, the oracle, the spec-oracle overlay) and write it to `<target>/.servo/plans/<spec-id>/plan.json`. This is the **last step of Servo Compile**: it decides nothing new, classifies nothing, and adds no gate — it only references what Compile already produced. Fire this skill when the user wants to: - "compile the execution plan for this spec" - "produce the Compile→Run handoff" - "emit `plan.json` for this spec" - "assemble the ADR-0016 plan for <spec>" / "wire up the Compile→Run artifact" Do NOT fire on: - "is this spec suitable for EDD?" / "what evidence is missing?" — that's `/servo:edd-suitability` (015). This skill *consumes* the suitability verdict as a hard precondition; it does not decide it. - "classify the acceptance criteria" / "build the spec-oracle overlay" —

Author a frozen UI design-fidelity eval component for a project — the non-deterministic sibling of servo's deterministic component templates. Use when a project must be built to match a design mockup ("does the UI match the design?") and that fidelity should drive or gate the loop. Captures app vs reference screenshots, judges fidelity with a pinned vision model (n-sampled, lower-bound), freezes the definition, and installs a `score_design_fidelity` component into the project's `oracle.sh`. Do not use for deterministic checks (use scaffold-init / spec-oracle) or for the per-iteration judge agent.

Decide whether an engineering spec is suitable for Evaluation-Driven Development before any unattended loop runs against it. Emits the ADR-0015 **suitability verdict** — a closed three-state gate (`suitable` / `needs_evidence` / `unsuitable`), fail-closed, with an actionable `missing_evidence` checklist — written to `<target>/.servo/suitability/<spec-id>.json`. A gate, not a score: the loop never optimizes toward a meaningless green oracle on work that cannot be evaluated. Fire this skill when the user wants to: - "is this spec/work suitable for EDD?" / "should this go through servo?" - "what evidence is this spec missing?" / "why isn't this ready to compile?" - "why won't Compile proceed on this spec?" / "explain the suitability verdict" - "re-check suitability after I added tests" / "is <spec> EDD-shaped yet?" Do NOT fire on: - "scaffold the oracle" / "install servo" / "detect signals" — that's `/servo:scaffold-init` (001). Suitability *reads* the signals it detects; it does not synthe

Run heartbeat or set up a scheduled heartbeat for a servo-scaffolded target. The heartbeat is servo's Routine-triggered front-end: discover project signals into a triage inbox, read the heartbeat inbox, dispatch heartbeat findings, or run heartbeat end to end under one whole-heartbeat cost ceiling. Fire this skill for requests mentioning a Routine, cron, GitHub Actions schedule, or a scheduled agent that should invoke servo on an interval. Do NOT fire on: - "run the oracle" / "score this code" / "oracle score" — that's `/servo:quality-gate`. - "iterate on this" / "run an agent loop" / "fix this manually with the loop" — that's `/servo:agent-loop`. - "set up servo" / "scaffold the oracle" / "install servo in this project" — that's `/servo:scaffold-init`. - "install the oracle hook" / "oracle hook status" — that's `/servo:oracle-hook`. - "spec oracle" / "turn this spec into checks" — that's `/servo:spec-oracle`. - "design eval" / "visual fidelity score" — that's `/servo:design-

Run the scaffolded `oracle.sh` against a target project and surface a closed-contract result (0 = pass / 1 = below threshold / 2 = env error). Also exposes an `audit` subcommand that prints the servo install manifest without invoking the oracle. Fire this skill when the user wants to: - "score this code" / "score this project" / "run quality gate" - "run the oracle" / "what's the oracle score?" - "check the oracle" / "did the oracle pass?" - "show me the gate result" / "is the gate passing?" - "what does this servo install include?" (→ audit subcommand) Do NOT fire on: - "set up servo" / "scaffold oracle" / "install servo" — those are `/servo:scaffold-init`'s territory (a sibling skill, not this one). - "fix the failing test" / "make tests pass" / "debug this test" — out of scope. The gate reports; it does not modify code or tests. - "run my tests" / "lint this" — call the underlying tool directly; the gate is the composite scorer, not a single-tool runner. - "review my code"

Compile a spec or slice into a reviewable, project-owned **evidence overlay**: map each acceptance criterion to a deterministic check family, write a `plan.md` + `checks.json` under the spec's own `oracle/<id>/` directory, compile the checks into an installable `oracle.sh` component, and freeze it behind an explicit approval so an unattended loop can optimize against the spec's promises instead of a hand-waved single judge. Fire this skill when the user wants to: - "generate an oracle for this spec" / "spec oracle for <spec>" - "turn this spec/slice into checks" / "compile acceptance criteria into checks" - "evaluate / specify an oracle for a spec" / "what would servo judge for this spec?" - "plan the evidence for <spec>" / "install the spec overlay" / "approve the spec oracle" Do NOT fire on: - "run the oracle" / "score this code" / "what's the oracle score?" — that's `/servo:quality-gate`. This skill *builds and installs* the overlay; quality-gate *runs* it. - "iterate on this" / "ru

Install, uninstall, or report the status of servo's meta-judge `Stop` hook in a target project. The installed hook scores every assistant turn against the scaffolded oracle (via `gate.py`) and blocks the stop with a structured retry hint when the oracle is below threshold — the interactive cousin of `/servo:agent-loop`. Installing it mutates `<target>/.claude/settings.json`, so it is a **Tier-2** surface: offered explicitly, never auto-installed. Fire this skill when the user wants to: - "install the oracle hook" / "add the Stop hook" / "set up the meta-judge" - "uninstall the oracle hook" / "remove the meta-judge" / "remove the Stop hook" - "is the oracle hook installed?" / "oracle hook status" / "check the hook" Do NOT fire on: - "run the oracle" / "score this code" / "what's the oracle score?" — that's `/servo:quality-gate` (a one-shot score, not the hook installer). - "iterate on this" / "run an agent loop" / "headless loop" — that's `/servo:agent-loop` (the headless cousin; this hoo