Install
$ agentstack add skill-assertchris-claude-skills-custom-workflow-implement ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Implement
Parameters (from ARGUMENTS)
| Key | Purpose | Default | |---|---|---| | doc-path: | Explicit path to the plan doc (skips feature doc discovery) | — | | impl-context: | Extra context prepended to each sub-agent implementation prompt | — | | pre-commit: | Shell commands to run before committing (newline or semicolon separated) | npm run check | | completion-via: | How to check phase completion: progress (Progress section in doc) or git-log (commit message starts with Phase N:) | progress |
Step 1: Resolve the plan doc path
Check $ARGUMENTS for a doc-path: key.
- If
doc-path:is present: use it directly. Skip feature doc discovery. - If
doc-path:is absent:
- Run
git branch --show-currentto get the branch name. - Find the feature doc in
features/whose filename starts with the branch number.
If on main and no doc-path: was given, warn the user and stop — implementation should run on a feature branch.
Step 2: Resolve parameters
impl-context:— extract from ARGUMENTS, or use empty string.pre-commit:— extract from ARGUMENTS, or default tonpm run check.completion-via:— extract from ARGUMENTS, or default toprogress.
Step 3: Find incomplete phases
Read the plan doc. It contains phases as headings (e.g. ### Phase 1 — Title or ### Phase 1: Title).
Determine which phases are already complete based on completion-via::
progress: The Progress section at the bottom of the doc tracks Completed / In Progress / To Do. Any phase listed under "Completed" is done — skip it.git-log: Rungit log --oneline. Any commit whose message starts withPhase N:means that phase is complete — skip it.
Step 4: Implement each incomplete phase
For each incomplete phase, in order:
- Launch a sub-agent (using the Agent tool) with the following prompt:
"[impl-context, if any]
You are implementing a single phase of a feature. Here is what to do:
Phase: [phase number and title from the plan doc]
Instructions:
- Read CLAUDE.md for pre-commit checks, code style rules, and testing conventions
- Read the plan doc at [path] for full context on this phase
- Implement ONLY this phase
- Run the following pre-commit checks and fix any issues: [pre-commit commands]
- [If completion-via is 'progress':] Update the plan doc's Progress section: move this phase from To Do/In Progress to Completed
- Commit all changes with a message starting with 'Phase N: ' (e.g. 'Phase 1: Add Question types and extend PlanState')
Do NOT work on any other phase. Stop after committing."
- After the sub-agent finishes, verify completion based on
completion-via::
progress: Re-read the plan doc's Progress section. Confirm this phase now appears under "Completed".git-log: Rungit log --oneline -5. Confirm the most recent commit starts withPhase N:for this phase.- If verification fails, tell the user which phase failed and stop — do not continue.
- Move to the next incomplete phase.
When all phases are done, tell the user: "All phases complete."
If every phase is already complete, report "All phases already complete."
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: assertchris
- Source: assertchris/claude-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.