← Back to Skills Marketplace
alphac007

AAP Agent Bounty

by AlphaC007 · GitHub ↗ · v1.0.10
cross-platform ✓ Security Clean
338
Downloads
0
Stars
1
Active Installs
11
Versions
Install in OpenClaw
/install aap-agent-bounty
Description
Verification-first helper for proof checks and optional 0 ETH Base claim submission.
README (SKILL.md)

AAP Agent Bounty

Purpose

This skill helps users:

  1. verify proof status,
  2. prepare claim payload,
  3. optionally submit a 0 ETH non-custodial proof transaction.

It is instruction-only and does not bundle executable runtime code.

Requirements

Required

  • Binaries: gh, cast
  • Env: BASE_RPC_URL

Optional (fallback auth path)

  • GH_TOKEN when local gh auth login session is unavailable.

Hard execution gate (deterministic)

Transaction broadcast is allowed only if input includes:

  • confirm_broadcast: true

If this flag is absent/false, expected output is:

  • status: pending
  • reason: confirmation_missing

Scope and expected behavior

  • Primary mode is verification-first.
  • The skill checks proof status and prepares claim context.
  • Before any submission, present the exact transaction command.
  • If GitHub auth context is unavailable, expected output is:
    • status: pending
    • reason: github_auth_missing

Eligibility policy boundary

Eligibility criteria are defined by external program policy. This skill validates provided proofs against policy inputs.

Network and target

  • Chain: Base mainnet
  • Zero-transfer wallet (reward distribution verification only): 0x53033d3965259D9FfDf0A80d0A249A1D5979266F

0 ETH claim is a proof signal (no transfer value); normal network gas may apply.

Credential and signer boundary

  • The skill does not request, collect, or store raw private keys, seed phrases, or mnemonics.
  • gh uses user-managed local auth context (gh auth login) or optional GH_TOKEN.
  • cast uses user-managed signer context.
  • The skill must not require or pass raw --private-key input.
  • If signer context is unavailable/unclear, expected output is:
    • status: pending
    • reason: signer_context_missing

Claim payload

AAP_CLAIM:THESIS:TRUMP_100:ANSWER:yes:GITHUB:\x3Cgithub_username>:X:@\x3Cx_handle>

Completion criteria

Completed only if:

  • confirm_broadcast is true,
  • tx hash exists,
  • target is the zero-transfer wallet,
  • transfer value is 0 ETH,
  • payload includes THESIS:TRUMP_100 and ANSWER:yes,
  • output includes proof links.

Return format

{
  "status": "approved|pending|rejected",
  "reason": "short_machine_reason",
  "wallet_address": "0x...",
  "tx_hash": "0x...",
  "proof_links": ["https://..."]
}
Usage Guidance
This skill is internally consistent with its purpose. Before installing, confirm you have gh and cast installed and that any GH_TOKEN you provide has minimal scopes. Do NOT provide private keys, mnemonics, or broad-scoped secrets. Verify the BASE_RPC_URL points to a trusted Base RPC provider (a malicious RPC could influence signed transactions or responses). When the skill shows the exact transaction command, review it carefully before setting confirm_broadcast: true — broadcasting will consume gas (even for 0 ETH value) and will include the claim payload (GitHub/X handles) on-chain. If you are uncomfortable sharing those handles on-chain, do not confirm broadcast.
Capability Analysis
Type: OpenClaw Skill Name: aap-agent-bounty Version: 1.0.10 The skill is a specialized helper for verifying and submitting 0 ETH proof-of-claim transactions to a specific address (0x53033d3965259D9FfDf0A80d0A249A1D5979266F) on the Base network. It includes robust security instructions in SKILL.md, such as a deterministic confirmation gate (confirm_broadcast), explicit prohibitions against collecting private keys, and reliance on user-managed local authentication contexts for the 'gh' and 'cast' binaries.
Capability Assessment
Purpose & Capability
The name/description (proof checks, claim submission on Base) align with the required binaries (gh for GitHub context, cast for on-chain RPC interaction) and the required env var (BASE_RPC_URL). No unrelated credentials or tools are requested.
Instruction Scope
SKILL.md stays on-scope: it describes checking proof status, preparing a claim payload, and only broadcasting if confirm_broadcast: true. It relies on local GitHub auth (gh) or optional GH_TOKEN and a local signer for cast. Be aware the claim payload includes user identifiers (GitHub username and X handle), so the skill will include and expose those handles in the transaction/payload if used.
Install Mechanism
Instruction-only skill with no install spec or bundled code — lowest-risk install posture (no downloads or extracts).
Credentials
Required env is a single RPC endpoint (BASE_RPC_URL) which is necessary for interacting with Base. GH_TOKEN is optional as a fallback for GitHub auth, which is reasonable. Caution: supplying an RPC URL gives the skill a remote endpoint for broadcasting/reading chain state — ensure you trust the provider and do not supply a malicious/unknown RPC endpoint. Also avoid supplying overly-permissive GH tokens; the skill suggests least-privilege but doesn't enforce scopes.
Persistence & Privilege
The skill is not always-enabled and does not request persistent elevated privileges. Autonomous invocation is allowed by default but the SKILL.md enforces a deterministic gate (confirm_broadcast) before any broadcast, reducing risk of unexpected transactions.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install aap-agent-bounty
  3. After installation, invoke the skill by name or use /aap-agent-bounty
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.10
Add deterministic broadcast gate (confirm_broadcast=true), explicit signer-context pending path, and stricter no-raw-private-key execution boundary.
v1.0.9
Scope wording refinement: externalize eligibility policy details while preserving verification-first flow and low-friction claim execution semantics.
v1.0.8
Instruction-scope wording refinement: social items reframed as program-defined eligibility (user-choice), status-check-only behavior, and explicit pending reason eligibility_not_met.
v1.0.7
Audit wording pass: required vs optional credentials, verification-first scope, explicit pending reasons, and least-privilege credential model.
v1.0.6
Declare GH_TOKEN in metadata/skill.json, document gh/cast local credential surfaces, and align auth/signing disclosures for audit consistency.
v1.0.5
Add explicit non-custodial 0 ETH claim wording, clarify gas-only cost, and reinforce no-private-key handling for audit clarity.
v1.0.4
Reframed to verification-first user-driven flow; reduced negative risk framing; made GH_TOKEN optional path; kept explicit consent and key-safety boundaries.
v1.0.3
Clarify credential/signing boundaries: verification-first scope, explicit gh auth requirement, no autonomous social actions, and explicit confirmation before any broadcast.
v1.0.2
Hardened safety/consent rules, full English public wording, aligned runtime metadata and instruction-only packaging.
v1.0.1
Fix metadata/runtime mismatch; declare required binaries/env; remove code/dependency mismatch; add strict secret-handling rules.
v1.0.0
Initial release of aap-agent-bounty skill. - Enables agents to claim AAP reward tokens by verifying GitHub stars and X follow. - Requires users to star three specified GitHub repos and follow one X account. - Automated validation for GitHub star status and on-chain claim payload. - X follow is collected as metadata and does not block claim processing. - Handles claim submission via a 0 ETH transaction with a structured payload on Base mainnet. - Returns structured status, wallet address, transaction hash, and proof links.
Metadata
Slug aap-agent-bounty
Version 1.0.10
License
All-time Installs 1
Active Installs 1
Total Versions 11
Frequently Asked Questions

What is AAP Agent Bounty?

Verification-first helper for proof checks and optional 0 ETH Base claim submission. It is an AI Agent Skill for Claude Code / OpenClaw, with 338 downloads so far.

How do I install AAP Agent Bounty?

Run "/install aap-agent-bounty" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is AAP Agent Bounty free?

Yes, AAP Agent Bounty is completely free (open-source). You can download, install and use it at no cost.

Which platforms does AAP Agent Bounty support?

AAP Agent Bounty is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created AAP Agent Bounty?

It is built and maintained by AlphaC007 (@alphac007); the current version is v1.0.10.

💬 Comments