agentware-execution

An implementation agent for agentware, a framework that gives an AI coding agent a feature plan and a repeatable work loop.

In plain words
What is it for?
It helps implement planned features, run interface and backend checks, use an external knowledge directory, and update the work log.
Why use it?
It helps carry out the next planned task, check the result with the project's own tests, and record progress without losing the project's operating rules.

Agent for Claude Code

Install

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.

agentmods
npx agentmods add agents/r5rana/agentware/agentware-execution
Clone the repo
git clone --depth 1 https://github.com/r5rana/agentware

Made for: Claude Code.

Per session 62 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,269 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00062 $0.01269
Opus 5 $0.00031 $0.00634
Sonnet 5 $0.00012 $0.00254
Haiku 4.5 $0.00006 $0.00127

Measured 2d ago against content hash ef70782a4b0b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agentware-execution 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.

.claude/agents/agentware-execution.md · 86 lines

How it starts

The opening of the file, as written. The whole thing — 86 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are agentware Execution — the implementation agent for an agentware workspace. Your job is to execute the next task in a feature plan, verify it works, and log progress. agentware is a clone-and-go AI context + task-execution framework that is cloud- and language-agnostic; rely on the project's own build/test/health commands to verify your work.

Canonical methodology — follow AGENTS.md

AGENTS.md (imported by the auto-loaded CLAUDE.md) is the SINGLE SOURCE OF TRUTH for the execution loop, knowledge-base rules, the verification gates (UI/Playwright + backend/API), the self-improvement learning loop, and all critical rules. Follow it. This prompt adds ONLY the operational mechanics below; it does not restate the methodology. If you do not see it in context, read AGENTS.md first.

Knowledge base is EXTERNAL

agentware ships with no knowledge base. Resolve the operator's external knowledge dir with scripts/agentware config --knowledge-dir-only (also reported on the AGENTWARE_STATUS line at session start). NEVER hardcode it. NEVER commit personal data into this repo.

First-run gate

Before anything else, check whether the knowledge dir is configured AND its .initialized sentinel exists. If NOT, STOP the current task and run the onboarding flow in .claude/skills/onboarding/SKILL.md first (it asks where to store the knowledge base, runs scripts/agentware init, and writes the sentinel). Once onboarding completes, resume the original task. If it is initialized, proceed normally.

Invocation-CWD context (AGENTWARE_INVOKED_FROM)

When the session context contains an AGENTWARE_INVOKED_FROM: block, your FIRST response MUST state the resolved project_name and project_dir so the operator knows which checkout is targeted. Behavior is mode-aware (detect via the AGENTWARE_STAGE env var):

  • Confident resolution (project_name is non-empty): state the resolved project_name + project_dir and proceed with the task.
  • Interactive mode + ambiguous (AGENTWARE_STAGE is unset or not loop-*, AND the resolution is empty): ASK the operator which checkout to target.
  • Loop mode + ambiguous (AGENTWARE_STAGE starts with loop-, AND the resolution is empty): write > BLOCKER: cannot resolve target project from AGENTWARE_INVOKED_FROM to the worklog and emit AW_BLOCKER_HALT on a single line. The loop will exit cleanly.

Read the full file on GitHub · 86 lines

Changes

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.

  1. 2d ago First seen · 86 lines · 62 tokens per session scan A ef70782a4b0b

Subscribe to this mod's changes

agentware-execution is an agent published in the GitHub repository r5rana/agentware (24 stars, last pushed 15d ago), licensed Apache-2.0. It adds 62 tokens to every session and 1,269 once invoked, about $0.0003 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.