dr-api-execution-bootstrap

Bootstrap and enforce fast direct API execution in a workspace. Use when you want an agent to run API calls directly in-session, avoid unnecessary subagents,...

Security Scan

The skill's instructions largely match its stated purpose (bootstrapping direct in-session API execution) but it instructs the agent to read tokens/secrets and to persist/patch workspace startup files while the manifest declares no required config paths or environment variables — an incoherence that could lead to unexpected file writes or live API calls.

Purpose & Capability

Name/description (make direct API execution the default) aligns with the instructions: inspect startup files, persist an execution policy, run preflight checks, and validate. However, the manifest declares no required config paths or env vars while the runtime instructions explicitly target workspace startup files (AGENTS.md, MEMORY.md) and check for auth/tokens, so the metadata understates the skill's footprint.

Instruction Scope

SKILL.md tells the agent to inspect and surgically patch workspace startup files and to validate by running a 'small real dev test' if safe. It also instructs checking 'auth/token availability' and 'app code / required secret availability' — actions that may require reading environment variables, secret files, or other workspace config not declared in the manifest. The instructions therefore can cause file writes and live network calls and access to secrets without those access paths being declared.

Install Mechanism

Instruction-only skill with no install spec and no code files — low install-time risk. Nothing is downloaded or extracted.

Credentials

The skill requests no env vars in metadata, yet runtime guidance explicitly expects to check tokens, app/function keys, and 'required secrets'. This is a mismatch: the skill will attempt to access credentials or secrets (for preflight and validation) even though none are declared. That increases the chance it will read sensitive data unexpectedly.

Persistence & Privilege

The skill instructs the agent to persist an enforcement policy into workspace startup files (AGENTS.md, MEMORY.md) and enforce defaults for future sessions. It does not set always:true, but it does request persistent changes to workspace configuration and a behavioral change for future agent runs — the user should expect lasting modifications to workspace bootstrap files.

What to consider before installing

This skill aims to make direct in-session API calls the default and will edit workspace startup files and run a validation test. Before installing: 1) Expect it to read tokens/secrets and to write AGENTS.md / MEMORY.md — review and back up those files. 2) Insist the agent show diffs of any file changes and require your approval before any live API calls or writes. 3) If you do not want any code that accesses secrets or modifies startup files, do not install. 4) Consider running it in an isolated/dev workspace first to observe behavior. 5) Note the metadata mismatch (no declared config/env but instructions expect to access secrets) — ask the publisher to clarify which files and credentials it will read and to add explicit declarations.

Use this as an installer and execution-enforcer skill.

When the user asks to apply this skill, apply it immediately. Do not ask whether to enforce the policy.

Apply to this workspace

1. Inspect workspace startup/default files.
2. Persist the execution policy in workspace bootstrap files (AGENTS.md, MEMORY.md, or equivalent).
3. Patch surgically and preserve unrelated instructions.
4. Validate with the strongest safe real test available.
5. Report either Configured and validated or Configured, but blocked by: <reason>.

For the concrete application checklist, read references/APPLY.md.

Enforcement contract

1) Execution policy

Set and enforce these defaults for future sessions:

• prefer direct in-session API execution
• do not spawn subagents unless the user explicitly asks
• default to fast mode single-run chain
• do one upfront preflight only:
  • auth/token availability
  • app code / function key / required secret availability
  • one sanity endpoint check
• after preflight passes, execute the full API chain continuously without unnecessary pauses

2) Communication policy

Set and enforce:

• keep responses concise
• do not narrate every API call unless the user asks
• for blocked execution, report the blocker briefly and precisely
• for write operations, show one concise batch preview and wait for approval before executing

3) Operational execution rules

Read references/EXECUTION-PLAYBOOK.md and follow it for:

• bulk reads
• bulk writes
• failure handling
• resume behavior
• verification strategy

4) API-specific guidance

If the workflow involves Ikara-style CRUD/integration/service-compliance APIs, also read references/IKARA-PATTERNS.md before executing.

5) Validation requirements

After applying the rules, immediately validate them.

If safe and permitted, run one small real dev test and confirm:

• direct API call path works
• no subagent was spawned
• preflight + full-chain behavior is active

If real execution is not possible, run the strongest safe validation available and state exactly what prevented full validation.

6) Limits

If permissions, secrets, or tool access are missing:

• do not pretend they were enabled
• do not claim success
• report exactly what is missing
• keep the enforced policy in startup files anyway, unless file-write access is blocked

Reuse on other agents

If the user wants to replicate this behavior elsewhere, read references/INSTALL.md and provide the installation command plus the recommended bootstrap prompt.