Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/bjornjee/agent-dashboard/implementnpx skills add bjornjee/agent-dashboard --skill implementgit clone --depth 1 https://github.com/bjornjee/agent-dashboardWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00023 | $0.02220 |
| Opus 5 | $0.00012 | $0.01110 |
| Sonnet 5 | $0.00005 | $0.00444 |
| Haiku 4.5 | $0.00002 | $0.00222 |
Grade A, and why
implement scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 165 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Run the dispatch loop on a worktree with an approved multi-phase plan.
Opt-in. Invoked after /agent-dashboard:feature (or a sibling task skill) probes the user at plan approval and the user picks "Hand off to /agent-dashboard:implement". Each phase dispatches to a fresh Agent() subagent, keeping the orchestrator session slim. Re-invoking on a partially-done worktree resumes from the first pending phase — no separate resume mode.
Instructions
Follow these phases in order. Each phase has a gate — do not proceed until the gate is satisfied.
If a dispatched phase touches browser UI, Playwright, dev-server ports, screenshots, or interactive Browser/Chrome inspection, include ../_shared/ui-automation.md in the subagent context and pass the resolved worktree-local UI resources explicitly.
Include ../_shared/verification-profiles.md in every subagent prompt so Verification profile names remain defined even when agent-dashboard is installed without external core rules.
Phase 1: Locate worktree + plan
-
Confirm cwd is a worktree. Run
git rev-parse --show-toplevel; the path must match.../worktrees/<app>/<name>. If not, halt: "Run this from inside a feature worktree, not the main checkout." -
Read
.feature-plan-pathfrom the worktree root. Its single line is the absolute path to the approved plan markdown:cat .feature-plan-path -
Fallback if the sentinel is missing (older
/agent-dashboard:featurerun, or deleted):ls -lt ~/.claude/plans/*.md | head -5- Show the top 3 candidates via
AskUserQuestion, including the current branch name in the prompt for context. User picks one or "Other" to type a path.
-
Wait for env setup. Check the worktree root for the env sentinels:
.env-setup-done→ proceed..env-setup-failed→ surface contents, halt.- Neither →
/agent-dashboard:featurePhase 2's background agent is still running. Wait.
Gate: Worktree confirmed, plan file path resolved, env setup complete.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 165 lines · 23 tokens per session scan A fe6a64c8b731
implement is a skill published in the GitHub repository bjornjee/agent-dashboard (21 stars, last pushed 1mo ago), licensed MIT. It adds 23 tokens to every session and 2,220 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
skill-creator
Create or improve codewhale skills. Use when the user wants a new skill, wants to update an existing skill, or needs guidance on when a skill should be a skill versus MCP, hooks, tools, or a plugin scaffold.
documents
Compatibility alias for the docx skill. Prefer loading docx.
skill-to-subagent
Turn an existing pi skill into an isolated subagent and wire it into a project's implementation pipeline. Decides fitness first, writes the bridge agent, routes the model, tunes context inheritance, and wires a spawn checkpoint. Use on "wrap this skill as a subagent", "turn X into a subagent", "should this be a…
agent-network
Multi-Agent group chat collaboration system inspired by DingTalk/Lark. Enables AI agents to chat in groups, @mention each other, assign tasks, make decisions via voting, and collaborate. Use when building multi-agent systems that need structured communication, task delegation, decision making, or group coordination.
add-announcement
Helps add announcement cards to the sidebar banner system. Use when adding changelog entries, feature announcements, updates, or promotional banners to the Agenta sidebar. Handles both simple changelog entries and complex custom banners.
update-llm-model-list
Audit and update the supported LLM model list in assets.py against litellm's registry (models.litellm.ai). Use when adding new models, pruning outdated ones, or verifying the list is correct.