Skip to content
Skills Library
AI & agents · Automation

specflow

/specflow

Use when the user asks Codex to create, inspect, validate, run, dynamically review, patch, pause, continue, or resume Specflow/Aflow agent workflows in the current repository.

Repository
aflow ↗
Licence
Apache-2.0
Last repository change
28.06.2026

Assessment and security

90/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content25 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
4
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 28.06.2026, without references to retired models or APIs.

Licence

Apache-2.0, otwarta: można używać i zmieniać, z zachowaniem licencji i informacji o zmianach.

What it does and when to use it

Use when the user asks Codex to create, inspect, validate, run, dynamically review, patch, pause, continue, or resume Specflow/Aflow agent workflows in the current repository.

Original author description

Use when the user asks Codex to create, inspect, validate, run, dynamically review, patch, pause, continue, or resume Specflow/Aflow agent workflows in the current repository.

How it works

  1. Workspace
  2. Workflow Authoring
  3. Agent Servers
  4. Running
  5. Dynamic Review
  6. pauseAfterRun ACP Conversations
  7. Permission, Elicitation, Sessions
Original SKILL.md headings · 136 lines, 903 words
SKILL.md preview
---
name: specflow
description: Use when the user asks Codex to create, inspect, validate, run, dynamically review, patch, pause, continue, or resume Specflow/Aflow agent workflows in the current repository.
---

# Specflow

Use the plugin-provided Specflow MCP tools for real workflow operations. Do not
shell out to `specflow run` when the user wants Aflow Agent parity; that command
is the standalone CLI runner and intentionally does not cover dynamic runtime
patching or `pauseAfterRun` ACP conversations.

## Workspace

- Work in the user's target repository, not in the plugin directory.
- Pass the repository path as `cwd` to MCP tools when it is known.
- If the user gives a `serverUrl`, pass it through. Otherwise let MCP locate or
  start the persistent workspace Specflow server.
- Treat returned `serverUrl` and `runId` as durable handles for follow-up tool
  calls.
- Once a Specflow MCP tool first returns a `serverUrl` in the current chat,
  proactively surface the Specflow UI once for that `serverUrl`: in Codex
  Desktop, try to open the right-side browser to that UI; otherwise try to open
  Chrome to the UI. Do not open duplicate tabs for the same `serverUrl` in the
  same chat. If no browser action is available, do not block. Always tell the
  user both the UI URL and the `/design` URL.

## Workflow Authoring

- Before creating a new workflow, read `references/create-workflow.md` and
  `references/workflow-authoring.md`.
- Before adapting an existing workflow, read
  `references/fork-adapt-workflow.md` and
  `references/workflow-authoring.md`.
- For small edits to an existing workflow, read
  `references/workflow-authoring.md` only when schema details are needed.
- List workflows with `specflow_list_workflows`.
- Read YAML with `specflow_read_workflow`.
- Before adapting an existing workflow, fork it with
  `specflow_fork_workflow_to_local`.
- Write complete workflow YAML with `specflow_write_workflow`; local drafts
  should stay in `agentflows-local` unless the user explicitly wants a shared
  workflow.
- Usually reference files already in the repository by relative path instead of
  importing them.
- Use `specflow_import_assets` rarely: only when the user explicitly wants to
  copy files from outside the repository into
  `.aflow/.specflow/agentflow/assets/...` so the workflow is durable,
  shareable, commit-friendly, or reproducible.
- Validate with `specflow_validate_workflow` before running.

## Agent Servers

- Use `specflow_list_agent_servers` before writing workflow session
  `agentServerId` values.
- Use `specflow_list_agent_registry` when the user wants to inspect available
  registry agents.
- Use `specflow_install_registry_agent` or `specflow_update_registry_agent` only
  when the user explicitly asks to add or update a registry-backed agent server.
- Use `specflow_remove_agent_server` only when the user explicitly asks to
  remove an agent server.
- Do not treat workflow authoring as permission to change agent server config.
- For custom/headless or non-registry agents, ask the user to configure the
  agent in Specflow UI; do not write raw agent server JSON.
- Agent authentication is handled in Specflow UI or Aflow. If prepare/start
  reports auth is required, ask the user to authenticate there and retry.
- Use `specflow_get_agent_capabilities` or
  `specflow_refresh_agent_capabilities` to inspect supported ACP modes,
  permissions, and config options before hard-coding them in workflow YAML.

## Running

- Prepare a run first with `specflow_prepare_run` when variables or auth may be
  missing.
- Ask the user for missing variables in chat, then call prepare/start again
  with `variableValues`.
- If prepare/start reports missing agent authentication, tell the user to open
  Specflow UI or Aflow and authenticate there, then retry. Codex does not drive
  terminal/TUI auth through this plugin.
- Before starting a run, ask the user to choose Normal run or Dynamic review
  unless they already specified the mode.
- Start normal or dynamic runs with `specflow_start_run`.
- For long-running workflows, use bounded waits. If a tool returns before the
  run finishes, keep the `runId` and poll with `specflow_get_run` or continue
  with the relevant run tool.
- Use `specflow_pause_run`, `specflow_play_run`, `specflow_interrupt_run`, and
  `specflow_stop_run` only for existing active runs. `pause` and `interrupt`
  can be resumed with `play` when the server reports the run can continue.
  `stop` is terminal for that run; use continuation, rerun, or a new run
  instead of trying to play it afterward.
- Use `specflow_rerun`, `specflow_delete_run`, and
  `specflow_save_run_best_practice` only when the user explicitly asks to rerun,
  delete, or preserve a successful runtime snapshot as a workflow.

## Dynamic Review

Dynamic review is Codex-driven checkpoint review, not user-driven stepping. It
pauses after activations so Codex can inspect the completed node text,
optionally patch only the current run snapshot, and continue without asking the
user at every checkpoint.

- Start with `dynamicReview: true`.
- Inspect with `specflow_get_run_checkpoint`.
- Patch only when there is clear evidence that future workflow behavior should
  change.
- Patch with structured `specflow_patch_run_graph` operations, not full YAML.
- Continue with `specflow_run_to_next_checkpoint`.
- Ask the user during dynamic review only for explicit `pauseAfterRun`,
  permission/elicitation, auth, missing variables, ambiguity, or high-risk
  actions.

## pauseAfterRun ACP Conversations

`pauseAfterRun` is a server-owned ACP paused-node session, not MCP elicitation.

When a run reports paused nodes:

1. Call `specflow_list_paused_nodes` if needed.
2. Use `specflow_prompt_paused_node` for one or more messages to the paused ACP
   node.
3. Use `specflow_continue_paused_node` to close the authored pause.

In dynamic mode, call `specflow_continue_paused_node` with `play: false`, then
call `specflow_run_to_next_checkpoint` so the next activation is armed with
dynamic pause behavior.

## Permission, Elicitation, Sessions

- Use `specflow_list_pending_interactions` and
  `specflow_respond_interaction` for ACP permission or elicitation requests.
- Ask the user before selecting permission options unless their instruction is
  already explicit.
- Use session tools to list, restore, prompt, cancel, or close ACP sessions.
- Use `specflow_get_native_resume_commands` or
  `specflow_get_agent_session_native_resume_command` when the user asks how to
  resume a recorded session in a native agent CLI. Trust only returned verified
  commands; if unavailable, use ACP restore/inspect or ask the user.
- Do not look for MCP auth tools. Agent authentication is handled in Specflow UI
  or Aflow, then the Codex run can be retried.
Files in this package (4)
  • SKILL.md 7 KB
  • references/create-workflow.md 2 KB
  • references/fork-adapt-workflow.md 2 KB
  • references/workflow-authoring.md 13 KB

Versions

RepositoryScoreLicence
This page · duguyifei/aflow90Apache-2.0
duguyifei/aflow90Apache-2.0

Database snapshot: 13.11.2026 · package version 566da2807fd4 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 95specflow-skill
    Generates SpecFlow BDD tests for C#/.NET with Gherkin feature files and step bindings. Use when user mentions "SpecFlow", "C# BDD", ".NET Gherkin", "[Binding]", "[Given]/[When]/[Then]". Triggers on: "SpecFlow", "C# BDD", ".NET BDD", "step bindings", "[Binding]".
  • 90specflow-audit
    Audits a Specflow story/ticket for spec-compliance and surgically uplifts it — adding only the missing executable contract sections (SQL/RLS, TypeScript interfaces, invariant codes, Gherkin, acceptance criteria, Definition of Done, data-testid coverage) — then runs a pre-flight…
  • 100specflow-loop-selector
    Choose the lightest Specflow workflow that covers an ask’s size, uncertainty and risk, then select a loop only when needed. Use for work triage, proportionate verification, loop selection, or explicit spec-build, feature-build and Gate D requests. A lightweight decision does not start a loop.
  • 100reqnroll-testing
    Configures Reqnroll (the canonical .NET BDD framework) - install via `dotnet add package Reqnroll`, author `.feature` files in Gherkin, write step bindings as `[Given/When/Then]`-decorated methods in any C# class, runs via `dotnet test`. Reqnroll is the SpecFlow successor…
  • 100specflow-reqnroll
    When the user wants to design, implement, debug, or migrate BDD tests on .NET using SpecFlow or its open-source successor Reqnroll. Use when the user mentions "SpecFlow," "Reqnroll," "[Binding]," "[Given]," "[When]," "[Then]," ".feature file in .NET," "SpecFlow+ LivingDoc,"…
  • 90agent-workflow
    Create, update, run, resume, or continue agent workflows. Use when the user asks to create or fix workflow JSON, execute an existing workflow, run a workflow with a prompt, resume a workflow run, or continue the next node in a waiting multi-step workflow.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.