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 commands/alexsds/ade-workflow/executegit clone --depth 1 https://github.com/alexsds/ade-workflowWhat 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 | $0.00012 | $0.00711 |
| Opus 5 | $0.00006 | $0.00356 |
| Sonnet 5 | $0.00002 | $0.00142 |
| Haiku 4.5 | $0.00001 | $0.00071 |
Grade A, and why
execute scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
You are executing the ADE build workflow.
Step 1: Find the approved plan
Look in .ade/docs/plans/ for the most recent plan file. Read it and verify its status is approved or draft (if the user explicitly says to proceed).
If no plan exists, tell the user to run /ade:plan first.
Step 2: Read project settings
Read .claude/ade.local.md for commit style configuration.
Step 3: Update plan status
Update the plan's frontmatter status from draft/approved to in-progress.
Step 4: Create the team
Use the TeamCreate tool:
{
"team_name": "ade-build",
"description": "Generator + Evaluator building from approved plan"
}
Step 5: Create tasks from the plan
Read the plan and create a task (using TaskCreate) for each feature/deliverable. These tasks form the shared task list the team works from.
Step 6: Spawn teammates
Spawn both teammates using the Agent tool with team_name: "ade-build":
Generator — spawn with:
name: "generator"team_name: "ade-build"subagent_type: use the ade-generator agentprompt: Tell it to read the plan, read${CLAUDE_PLUGIN_ROOT}/skills/ade-generation/SKILL.mdfor methodology, check the task list, claim the first feature task, implement it, self-verify, commit to git, then message the evaluator via SendMessage (to: "evaluator") with: feature name, files changed, how to test. After evaluator responds, iterate if needed. When approved, mark task complete, claim next task.
Evaluator — spawn with:
name: "evaluator"team_name: "ade-build"subagent_type: use the ade-evaluator agentprompt: Tell it to read${CLAUDE_PLUGIN_ROOT}/skills/ade-evaluation/SKILL.mdfor methodology, then wait for messages from the generator. When a feature is submitted, load rubrics from${CLAUDE_PLUGIN_ROOT}/rubrics/and.ade/rubrics/, load testing tools from${CLAUDE_PLUGIN_ROOT}/testing-tools/and.ade/testing-tools/, test the feature, score against rubric criteria, then message the generator via SendMessage (to: "generator") with the scored report. If all criteria pass, say APPROVED. If any fail, say FAILED with specific fixes.
Step 7: Monitor progress
The team communicates via SendMessage. The workflow per feature:
- Generator implements feature, commits, messages evaluator
- Evaluator tests, scores, messages generator with pass/fail
- If failed: Generator iterates, re-submits
- If passed: Generator marks task complete, moves to next feature
When all tasks are complete, send teammates a shutdown message:
SendMessage({ to: "generator", message: { type: "shutdown_request" } })
SendMessage({ to: "evaluator", message: { type: "shutdown_request" } })
Then notify the user that the build is complete and suggest running /ade:done to archive the plan.
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.
- yesterday First seen · 66 lines · 12 tokens per session scan A 8617cbbf460e
execute is a command published in the GitHub repository alexsds/ade-workflow (1 stars, last pushed 5mo ago), licensed MIT. It adds 12 tokens to every session and 711 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 2026-08-31.
Other commands, from other repositories
feature
Orchestrate a complete feature through discovery, spec, implementation, and review.
test
Design or run focused test validation for a task, bug, or diff.
mvp-spec
Research and produce a strict MVP spec with small 1-2 hour tasks and explicit out of scope.
research
Research a technical or product question.
goal
Set a goal and enter the RALPH loop — keep working autonomously until the goal is genuinely done or the user stops you. Use /goal "objective" to start, /goal status to check, /goal complete to finish, /goal clear to stop.
selfloop
Start or control a persistent SIPS loop whose only objective is improving SIPS and the agent operating it through measured, verified iterations.