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/dcc-mcp/dcc-mcp-renderdoc/renderdoc-capturenpx skills add dcc-mcp/dcc-mcp-renderdoc --skill renderdoc-capturegit clone --depth 1 https://github.com/dcc-mcp/dcc-mcp-renderdocWhat 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 | $0.00044 | $0.00564 |
| Opus 5 | $0.00022 | $0.00282 |
| Sonnet 5 | $0.00009 | $0.00113 |
| Haiku 4.5 | $0.00004 | $0.00056 |
Grade A, and why
renderdoc-capture 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 3d 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.
What it actually says
RenderDoc Capture
Call get_version before capture. capture_program starts only the explicit executable and
arguments, never a shell. Set trigger_after_secs to request one capture through RenderDoc's
official local Target Control API. This headless trigger requires qrenderdoc beside
renderdoccmd; it does not focus a window or synthesize keyboard input. hook_children remains a
RenderDoc launch option. Set trigger_process_name only with hook_children=true; the sidecar
checks the launched target first and, if its name differs, follows only its official NewChild
messages to find a unique matching child. A child name without child hooking fails before launch.
The official RenderDoc runtime supports Windows and Linux, not macOS (where this project runs
Python unit tests only). On headless Linux, run the adapter under Xvfb or provide another working
X/Wayland display; the official archive does not include Qt's offscreen platform plugin.
Capture tools inspect each new .rdc before reporting success. A capture with no frame work such
as Draw, Dispatch, Clear, Copy, Resolve, Blit, render-pass work, or command-list execution is
rejected with chunk diagnostics while the .rdc remains at the requested output path; retry while
the intended target is actively rendering or correct the selected child process.
Use capture_process when Steam or another platform client must launch the target first. It
injects into the explicit PID, connects to the returned Target Control ident, requests one capture
after the configured delay, and waits for the resulting file. It never takes foreground focus or
emits a hotkey. Inject before the target creates its graphics device; late injection cannot recover
resources that RenderDoc did not observe. Prefer capture_program when the executable can be
launched directly. Neither tool terminates the target process.
What ships with it
5 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.
- 3d ago First seen · 46 lines · 44 tokens per session scan A 153f0fd71531
renderdoc-capture is a skill published in the GitHub repository dcc-mcp/dcc-mcp-renderdoc (1 stars, last pushed 9d ago), licensed MIT. It adds 44 tokens to every session and 564 once invoked, about $0.0002 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 2026-08-31.
Other skills, from other repositories
godot-runtime
Domain skill — Return game-peer connection and play status. Execute one strict project-manifest action for bounded playtest control. Return one bounded page of the running game hierarchy. Return one bounded property page from a game node. Set a property on a running-game node. Compatibility-only broad public-method…
godot-editor
Domain skill — Return errors captured by the DCC-MCP editor plugin. Capture the selected editor viewport to a PNG. Capture game pixels on the runtime thread and encode the PNG in the adapter. Run one @tool method with an observational budget and optional caller-driven chunks. Clear errors captured by the DCC-MCP…
godot-export
Domain skill — Inspect export presets and templates, package Godot projects for desktop, Web, and mobile targets, and validate packaged-only resource, font, signing, and hosting failures. Use for export planning and release preflight, not for editing gameplay or scene content.
godot-animation-tree
Domain skill — Create and configure an AnimationTree node. Return AnimationTree parameters and tree type. Set an AnimationTree parameter. Add a state-machine animation node. Remove a state-machine node. Add a state-machine transition. Remove a state-machine transition. Add or replace an AnimationNode in a blend tree.
godot-batch-refactor
Domain skill — Find edited-scene nodes by class. Return scene signal connections. Set a property on matching edited-scene nodes. Search project text for a node path or name. Return dependencies for a scene resource. Update matching nodes across project scenes. Find project references to a script or resource. Detect…
godot-node
Domain skill — Add a typed node with initial properties. Delete a node with undo support. Duplicate a node and its owned children. Reparent a node while preserving ownership. Set an existing node property. Return stored and editable node properties. Create and assign a Resource to a node property. Apply a Control…