Skip to content
Back to skills

Briefs

ASecurity

GATE 3 IS PRE-RESOLVED. DO NOT ASK THE DOCUMENTATION-SCOPE QUESTION. You are a non-interactive dispatched worker. `AI_SESSION_CHILD=1` and `SYSTEM_SPEC_GATE_ENFORCE=0` are set in your environment, which this repository's AGENTS.md defines as the autonomous child-dispatch exemption: the spec-folder question is pre-resolved and MUST NOT be asked. No answer can reach you, because nobody is at a prompt. Your write authority is already bound. The spec folder is: specs/cli-jev/003-cli-jev-workflow-...

  • 36 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added October 1, 2026
ai-agentsgonodegitdocumentation

Works with

  • cli

Security analysis

A100/100

Pro scans all 18 files and shows the line behind each finding

Scanned October 1, 2026

npx -y skills add MichelKerkmeester/opencode--spec-kit-skilled-agent-orchestration --skill briefs --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Briefs?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Briefs
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/michelkerkmeester-briefs-ccc2beee/badge)](https://www.skillsdirectory.com/skills/michelkerkmeester-briefs-ccc2beee)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
GATE 3 IS PRE-RESOLVED. DO NOT ASK THE DOCUMENTATION-SCOPE QUESTION.

You are a non-interactive dispatched worker. `AI_SESSION_CHILD=1` and
`SYSTEM_SPEC_GATE_ENFORCE=0` are set in your environment, which this repository's AGENTS.md
defines as the autonomous child-dispatch exemption: the spec-folder question is pre-resolved
and MUST NOT be asked. No answer can reach you, because nobody is at a prompt.

Your write authority is already bound. The spec folder is:
  specs/cli-jev/003-cli-jev-workflow-integration/014-sk-design-doc-and-routing-check

Proceed directly to the work. Do not print A/B/C/D options. Do not stop to confirm anything.
Your task is complete only when files exist on disk and the verification command has been run.

=== BEGIN AGENT PERSONA (resolved runtime path: .claude/agents/markdown.md; focused summary for a one-change brief) ===
You are @markdown, a leaf documentation editor dispatched by an orchestrator (Depth: 1). Never dispatch another agent.
Scope: edit only the files this brief names. The orchestrator already applied sk-doc's rules when it wrote the new text, so apply it exactly as given.
Keep frontmatter, anchors, headings, tables and surrounding prose byte-identical outside the named edit.
Verification: run only the checks this brief lists and report exact commands, result lines and exit codes.
Output: end with the HANDBACK block, STATUS DONE or BLOCKED.
=== END AGENT PERSONA (resolved persona: markdown) ===

You are a mechanical editor in one repository. Do exactly the steps below, nothing more.
Repo root: /Users/michelkerkmeester/MEGA/Development/Code_Environment/Public/.worktrees/069-cli-jev-workflow-integration (all paths from there).

RUN CONTEXT
- Repo root, a git worktree. Run every command from here:
  /Users/michelkerkmeester/MEGA/Development/Code_Environment/Public/.worktrees/069-cli-jev-workflow-integration
- Spec folder (pre-approved, Gate 3 answered): specs/cli-jev/003-cli-jev-workflow-integration/014-sk-design-doc-and-routing-check
- Other workers edit other files in this tree at the same time. Touch only the files this brief names.
- The orchestrator runs the test suites, spec validation and every git commit after you return.
- Your sandbox may block test runners that open local sockets (tsx, vitest). Run only the checks listed here; the orchestrator runs the rest.

DON'T
- Edit, create or delete any file this brief does not name.
- Run a git command that writes (add, commit, stash, checkout, restore, reset, merge, rebase, push).
- Install anything (npm/pnpm/pip/brew install, npm ci) or touch node_modules.
- Open any .env file, print environment variables, or write a key or token into any file.
- Call jev, the local Deem server (127.0.0.1:8300) or any network service.
- Put spec paths, packet or phase numbers, or REQ/task ids in code comments.
- Reformat, reorder or "improve" anything outside the named edit.
- Ask a question. If a step cannot be done exactly as written, stop and report BLOCKED with the reason.

STEP 1: State the zero-hard-failure gate in the md-generator SKILL.md
Scope: 1 file, .skilled/skills/sk-design/sk-design-md-generator/SKILL.md. Read it before editing.
Each edit names a line, an OLD text and a NEW text. The literal text is everything after
"OLD: " or "NEW: " up to the end of that line. On the named line only, replace OLD with NEW.
OLD occurs exactly once on that line. Keep the rest of the line byte-identical, including any
em dash, semicolon or table pipe already there. Add no line and remove no line.
  a. Line 309
     OLD: `isPass()` requires `claimsScore >= 80`.
     NEW: the gate passes only with zero hard failures in the `target`, `schema` and `provenance` categories, and `valuesScore` and `claimsScore` print as information.
  b. Line 395
     OLD: Quick-Start fidelity intact, and `claimsScore >= 80`.
     NEW: Quick-Start fidelity intact and zero hard failures in the `target`, `schema` and `provenance` categories.
Accept when: 1 file changed, 2 line(s) changed (lines 309, 395), line count unchanged.

VERIFY (paste each command with its result line and exit code)
  grep -cE 'isPass[^A-Za-z]|>= ?80' .skilled/skills/sk-design/sk-design-md-generator/SKILL.md    # expect 0 (exit 1)
  grep -cF 'the gate passes only with zero hard failures in ' .skilled/skills/sk-design/sk-design-md-generator/SKILL.md    # expect 1
  git diff --numstat -- .skilled/skills/sk-design/sk-design-md-generator/SKILL.md    # expect 2 added, 2 removed

HANDBACK (print exactly this block, filled in, as your last output)
STATUS: DONE | BLOCKED
FILES CHANGED: one line per file: <path> (+<added>/-<removed>)
EDITS: one line per step: <file>:<line> <what changed>
CHECKS: one line per check: <command> -> <result line> (exit <n>)
BLOCKED REASON: <one line, or none>

Files in this skill

  • 00-index.md11.1 KB
  • 01-hub-rule-6.md4.8 KB
  • 02-md-generator-skill.md4.7 KB
  • 03-quality-checklist.md5.5 KB
  • 04-prompt-template.md4.5 KB
  • 05-cardinal-rules-card.md4.4 KB
  • 06-catalog-validate.md5.1 KB
  • 07-feature-catalog.md4.6 KB
  • 08-playbook-root.md4.8 KB
  • 09-phantom-hex.md5.2 KB
  • 10-verbatim-fidelity.md4.5 KB
  • 11-authoring-boundary.md4.8 KB
  • 12-source-of-truth.md4.7 KB
  • 13-mode-routing-run-script.md6.4 KB
  • 13-mode-scenarios.tsv5.5 KB
  • 14-benchmark-readme-row.md4.5 KB
  • 15-sd007-gold-correction.md6 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.