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.
npx agentmods add skills/filliptm/comfyui_fl-mcp/workflow-assistantnpx skills add filliptm/ComfyUI_FL-MCP --skill workflow-assistantgit clone --depth 1 https://github.com/filliptm/ComfyUI_FL-MCPWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/filliptm/comfyui_fl-mcp/workflow-assistant)<a href="https://agentmods.dev/skills/filliptm/comfyui_fl-mcp/workflow-assistant"><img src="https://agentmods.dev/badge/skills/filliptm/comfyui_fl-mcp/workflow-assistant.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00051 | $0.02878 |
| Opus 5 | $0.00026 | $0.01439 |
| Sonnet 5 | $0.00010 | $0.00576 |
| Haiku 4.5 | $0.00005 | $0.00288 |
Grade A, and why
workflow-assistant 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 6d 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.
How it starts
The opening of the file, as written. The whole thing — 111 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Workflow Assistant
Use this skill when a user wants help with a live ComfyUI workflow, saved workflow JSON, generated image metadata, custom node graph, or FL-MCP tool session.
This skill is workflow-first. It borrows the old Ren Agent's operating discipline and cognitive modes, but does not adopt a persona, product UI, hosted chat runtime, or ren:// links.
Core Loop
- Compile the whole requested graph change once.
- For a whole-branch request such as find, jump, compare, clone, replace, or remove, call
workflow_branches_discoverfirst. Use exact returned workflow, graph, catalog, scope, branch, node, slot, and boundary-edge facts; never authorize a branch from its label or fingerprint. - Resolve comparison operands to two exact branch IDs under the same pinned catalog. One bounded listing may identify both uniquely; otherwise perform separate pinned discovery resolutions. Never guess two branches from one ambiguous text query.
- For whole-branch clone, replace, or remove, pass the exact discovery pins and mappings to
compile_workflow_branch_operation. Root operations return GraphPatch v2; supported nested operations return scoped GraphPatch v3. - For a new workflow, an added branch, a rewire, a removal, or another explicit topology change, call
compile_workflow_refinement_specfirst. Mask pixels and prompt text use their dedicated narrow lanes below; do not send value-only prompt edits through GraphPatch. - Include every requested role or exact class, value, edge, update/removal, attachment, and layout hint in one semantic request.
- Describe the desired endpoints rather than guessing intermediary classes or dynamic dotted paths. The compiler prefers direct compatibility, then may infer a unique bounded supported local converter route. Partner/API/heavy/output nodes require explicit intent.
- When the user says exactly, only, or no extra nodes, set
allow_inferred_converters=false. - Use deterministic existing-node selectors. Set
selected=truewhen the user says “selected”, “this”, or “that”; the compiler reads the live selection internally. - Do not precede the compiler with workflow JSON, overview, selection, node search/details, slots, values, or layout reads. It reads the active graph (including an empty canvas) and refreshed local native, partner, and custom schemas internally.
- For a whole-branch request such as find, jump, compare, clone, replace, or remove, call
- Respect deterministic outcomes.
- If
valid=true, pass the returnedapply_requestunchanged toapply_workflow_graph_patch. - After a successful whole-branch apply, call
resolve_workflow_branch_successorwith the unchanged apply envelope, compilerpending_successor_locator, and the apply result's exact aliases, workflow identity, and final graph hash. Treat its per-scope successor list as authoritative; zero or multiple successors are valid outcomes when explicitly attested. - If
needs_choice=true, show the ranked node, endpoint, or route candidates and wait for the user. Never accept an alphabetical tie-break. - If a schema is classified unsupported, stop and report it. Use lower-level schema tools only in a focused diagnostic follow-up, never to bypass the failed atomic route in the same run.
- If
- Apply one atomic GraphPatch.
- Keep the same application ID for retries so the idempotency ledger cannot duplicate work.
- Root GraphPatch v2 and scoped GraphPatch v3 may create, update, connect, disconnect, remove, attach, and lay out arbitrary acyclic branches with fan-in, fan-out, multiple sinks, dynamic inputs, and explicit widget-to-input conversion.
- It pins workflow, graph, catalog, and schemas; preserves unrelated state; visibly applies mutations in deterministic order; verifies the exact final graph; fully rolls back on failure; and never queues.
- Do not fall back to
create_nodes,set_node_values,connect_nodes_batch,remove_nodes, or the legacycompile_workflow_spec/resolve_workflow_spec/plan_workflow/apply_workflow_plancompilers after a GraphPatch validation/application failure.
- Trust exact verification.
- A successful application already verifies the graph, node facts, attachments, layout, and unrelated workflow envelope. Do not add redundant overview/JSON/slot/value/layout reads.
- Inspect again only when the result reports a mismatch, the user explicitly asks for a visual check, or a later execution/output review requires it.
- Queue only after validation.
- Check required sampler, decoder, loader, and save connections.
- After queueing, inspect queue/history/errors before declaring success.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 6d ago First seen · 111 lines · 51 tokens per session scan A 8547ba65d02f
workflow-assistant is a skill published in the GitHub repository filliptm/ComfyUI_FL-MCP (118 stars, last pushed 18d ago), licensed MIT. It adds 51 tokens to every session and 2,878 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.
Other skills, from other repositories
webgl-holographic-foil
A self-contained WebGL2 hero: thin-film interference over a crushed-foil surface whose palette shifts with the viewing angle; move the cursor to tilt the film.
general-video
Author or edit a custom HyperFrames composition when no specialized workflow fits, or when BRIEF.md sets flow: companion. Use for longer or multi-scene pieces, brand and sizzle reels, montages, static loops, static title cards, footage remixes, and freeform builds. Use motion-graphics instead for a short unnarrated…
html-ppt-hermes-cyber-terminal
OpenDesign + BYOK: choosing and wiring your own model, hands-on — cost, quality, and the routing decision. Built as a decision-grade AI literacy deck for engineers, IT, applied-AI teams.
html-ppt-taste-brutalist
16:9 HTML deck in tactical-telemetry / CRT-terminal taste. Deactivated-CRT charcoal slides, white-phosphor monospace, hazard-red accent, scanline overlay, ASCII syntax, density over decoration. Distilled from Leonxlnx/taste-skill brutalist-skill (Tactical Telemetry mode).
diagnostic-stem-delivery
Audio production with diagnostic analysis, timecode parsing from documents, and verified export workflow.
chengfeng-check-updates
剪辑环境的唯一管理者:就绪检查(skills 是否最新 → Runtime 是否配套)、Skills 更新激活、Runtime 安装与体检。用户说检查更新、安装剪辑环境、装播放器、检查剪辑环境、剪辑环境就绪了吗、配置转录凭证时使用;业务 Skill(剪口播/字幕/画面/导出)第 0 步也引用本 Skill 的就绪检查。不用于剪辑、字幕、画面、导出本身或项目数据迁移。.