Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    NightRaven

    1

    Pick up your repo’s roadmap with clear scope and test evidence. NightRaven helps your coding agent resume approved tasks, stay within the files you authorize, and report which checks ran and what still needs attention.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    NightRaven

    NightRaven

    Example session with this skill installed

    In the fixture, complete the approved roadmap using the NightRaven skill. Use available parallel agents, record worthwhile recommendations as tasks, and show the verification and handoff.

    • Read your context and instructions
    • Compiled the nightraven

    Result from a controlled NightRaven 1.3.0 trial

    Two coding agents worked in separate copies of a disposable Git project. The coordinator integrated their proposals one at a time and reran previously accepted checks after integration.

    TaskVerificationOutcome
    Show the currency on rendered totals2 of 2 checks passedAccepted on attempt 1
    Trim boundary whitespace from public labels3 of 3 checks passed, including the accepted totals checkAccepted on attempt 1 after independent review

    Independent review

    A separate read-only agent checked the integrated label change against the acceptance criteria and reviewed it for regressions. Both verdicts passed, and an independent rerun of the combined test suite passed 4 of 4 tests.

    Backlog and handoff

    The reviewer suggested adding documentation examples for tabs, line breaks, internal spaces, and blank input. NightRaven added that evidence and an acceptance check to the existing documentation task. The backlog kept one entry, linked it to the completed label change, and recorded that implementation awaits approval. The handoff recorded both accepted tasks, their checks, and the remaining decision. Both integrations recorded zero scope violations.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Give your coding agent a clear way to pick up an existing roadmap. NightRaven helps it resume approved work, respect the file scope you set, and hand back the checks behind each result.

    • Get independent review before accepting higher-impact changes. On supported coding hosts, a separate read-only agent checks the change against your acceptance criteria and looks for regressions the tests may have missed.
    • Turn suggestions into backlog tasks automatically. During authorized work, NightRaven records actionable suggestions with evidence and a way to verify them. If a matching task already exists, it updates that task. Ready follow-ups can start when your roadmap or standing approval covers them.
    • Keep work moving. See what needs attention, resume interrupted tasks, and continue independent work when something is blocked. Run tasks in parallel when your coding host supports isolated workers.

    Verify the result

    NightRaven records the allowed files, acceptance criteria, and tests that protect existing behavior. The included Python checker compares file changes against that scope and runs your declared commands. Your agent reports what passed, which revision was checked, and what still needs attention.

    What you’ll need

    Use an Agensi-capable coding agent with project read access. Implementation and backlog updates need write access, and automated verification needs Python 3.9+ and your project’s test commands. Independent review and parallel execution depend on the tools available in your coding host.

    • Reliability depends on the coding agent, available tools, project tests and user-defined acceptance criteria.
    • The portable checker verifies captured scope/protected evidence and declared commands when run. It does not sandbox edits, intercept every write or make editable baseline data tamper-proof.
    • Task selection, authorization, dependencies, attempt counters and semantic acceptance remain agent responsibilities.
    • Unlimited attempts and queue continuation do not remove no-progress blockers, unavailable prerequisites or agent/model/tool/platform resource limits.
    • Passing checks does not prove the absence of bugs or security issues.
    • Automatic checking needs Python 3.9+ and project tools. Without suitable tools, planning/manual review only and execution checks remain unperformed.
    • No model access, project dependencies, hosted service or autonomous background execution is included.
    • Human acceptance and device, deployment or production checks remain with the project owner.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 3 days ago

    Requires

    Python 3.9+
    Agensi-capable coding agent
    Project file tools
    Command execution
    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed6 days ago
    Updated3 days ago

    What's inside

    Frequently Asked Questions

    No. The text skill and standalone checker are self-contained. Your agent supplies file/command tools; the project supplies its runtime, dependencies and checks.

    The delivered Python checker runs explicitly declared project commands and compares captured file scope and protected evidence when your agent invokes it. It is not a filesystem sandbox or a background enforcer. The agent retains task policy and acceptance decisions.

    No numeric attempt ceiling is imposed on new tasks, and there is no fixed batch size for an authorized queue. A user-selected finite limit is respected. Repeated failure without relevant progress or an unchanged unavailable prerequisite requires diagnosis and a blocker; existing histories and accepted work stay protected.

    It can help write a task agreement, specification and check plan from supplied material. It must leave edits and execution checks marked unperformed until suitable tools are available. Python 3.9+ is needed for the automatic checker.

    It starts with existing project records and respects the owner's instructions. It does not require another memory bank or activate upstream bootstrap behavior.

    The skill grants no such authority. It does not require credentials or publication access; existing user authorization and tool controls continue to apply.

    New in Agents & Orchestration