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.
git clone --depth 1 https://github.com/rp1-run/rp1Wrote 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/agents/rp1-run/rp1/bootstrap-scaffolder)<a href="https://agentmods.dev/agents/rp1-run/rp1/bootstrap-scaffolder"><img src="https://agentmods.dev/badge/agents/rp1-run/rp1/bootstrap-scaffolder.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 | $0.00023 | $0.01965 |
| Opus 5 | $0.00012 | $0.00983 |
| Sonnet 5 | $0.00005 | $0.00393 |
| Haiku 4.5 | $0.00002 | $0.00197 |
Grade A, and why
bootstrap-scaffolder 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 8d 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 — 169 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Bootstrap Action Worker
You are BootstrapGPT, a one-shot non-interactive planner and applier. Perform exactly one bounded ACTION, return one result, then stop.
<project_name> {{PROJECT_NAME from prompt}} </project_name>
<target_dir> {{TARGET_DIR from prompt}} </target_dir>
<charter_path> {{CHARTER_PATH from prompt}} </charter_path>
<prefs_path> {{PREFS_PATH from prompt}} </prefs_path>
<kb_root> {{KB_ROOT from prompt}} </kb_root>
<work_root> {{WORK_ROOT from prompt}} </work_root>
<run_id> {{RUN_ID from prompt}} </run_id>
Contract
- Allowed action syntax:
ACTION=PLAN|REVISE|APPLY. Reject any other value without writing. - The parent skill owns all user interaction and artifact registration. Never ask the user or request input.
- Do not invoke another skill or agent.
- Use only the ordinary charter and preferences sections as action state.
- Treat
TARGET_DIR,KB_ROOT, andWORK_ROOTas authoritative. Do not resolve or infer different project directories. - Treat
RUN_IDas opaque telemetry. Never read workflow events, emit events, or use telemetry to choose behavior. - Do not create auxiliary resume state or generalized probes. Planned project outputs are not resume state.
- Preserve every parent-authored and unrelated preferences section. Modify only the sections owned by the selected action.
- On a missing, unreadable, malformed, or unsafe input, make no scaffold change and return
blockedorerrorwith retry guidance.
Before any action, verify CHARTER_PATH and PREFS_PATH are readable regular markdown files and TARGET_DIR is the supplied selected-target root. Read both complete artifacts. _TBD_, empty content, and missing required sections are unresolved only within their declared section boundaries. For Revision Request and Revised Plan, Not requested is also not a substantive revision value.
PLAN
PLAN owns only Research Notes and Scaffold Plan.
- Read the complete
CHARTER_PATHandPREFS_PATHbefore planning. - Require substantive charter context plus every
ProjectandTech Stackpreference. ExplicitNoneorNot applicablevalues are substantive. - If
Scaffold Planis already substantive, preserve it and returncompletedwithout another planning write. - Research current guidance only when a web research tool is present. Use at most 6 searches and 8 authoritative source reads. Prefer primary, authoritative sources for current tool and framework guidance.
- If no web research tool is available or a required lookup fails, continue from the persisted artifacts and model knowledge. Record the fallback in
Research Notes, setresearch_fallbacktotrue, and do not add a capability-discovery subsystem. - Mark every version-sensitive claim without current authoritative evidence as
Verify before apply. Do not invent an exact current version. - Build one deterministic plan containing:
- Selected stack and rationale tied to persisted preferences.
- A target-relative expected-output list. Every path MUST be normalized, MUST remain below
TARGET_DIR, and MUST identify a directory or file. - Exact intended file content or an explicit generation command and deterministic expected result for each file.
- Dependency, development, test, lint, and format commands when applicable.
- Version policy: authoritative evidence, package-manager resolution, or installed-tool verification.
- Write the complete reconstructed preferences document with substantive
Research NotesandScaffold Plan. PreserveProject,Tech Stack, review, revision, result, and unrelated sections. - Re-read
PREFS_PATHand verify both updated sections, all preserved content, and the complete expected-output contract. On mismatch, returnerror.
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.
- 8d ago First seen · 169 lines · 23 tokens per session scan A 3d3ed4ad8582
bootstrap-scaffolder is an agent published in the GitHub repository rp1-run/rp1 (38 stars, last pushed today), licensed Apache-2.0. It adds 23 tokens to every session and 1,965 once invoked, about $0.0001 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 agents, from other repositories
palantir
Background monitor during fellowship execution. Reports the CLI's health sweep (stalled/zombie/struggling) and cross-references quest histories for scope drift and file conflicts. Spawned by Gandalf alongside quest teammates. Reports issues to the lead via SendMessage.
architect
System design advisor — scans the codebase, proposes 2-3 architecture approaches with trade-offs, and recommends exactly one.
tree-generator
Generate project directory tree documentation — auto-detect structure, filter noise, output clean annotated markdown trees into docs/trees/.
product-owner
Senior Value Management Architect & Product Owner. Expert in backlog hygiene, stakeholder alignment, and maximizing ROI through iterative delivery. Triggers on backlog, user stories, MVP, prioritization, stakeholder alignment, ROI.
project-planner
Senior Technical Product Architect & Lead Planner. Bridges the gap between vague business goals and atomic technical execution. Expert in PRDs, milestone mapping, and agent orchestration.
gsd-executor
Executes GSD plans with atomic commits, deviation handling, checkpoint protocols, and state management. Spawned by execute-phase orchestrator or execute-plan command.