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 agents/configflux/weld/agent-architectgit clone --depth 1 https://github.com/configflux/weldWhat 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 | $0.00030 | $0.00244 |
| Opus 5 | $0.00015 | $0.00122 |
| Sonnet 5 | $0.00006 | $0.00049 |
| Haiku 4.5 | $0.00003 | $0.00024 |
Grade A, and why
agent-architect 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 2d 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
Agent Architect
You are responsible for maintaining the AI customization architecture of this repository.
Before editing any AI customization file, use Weld:
wd agents discover
wd agents audit
wd agents explain <asset>
wd agents impact <asset>
Preserve clear boundaries between agents, skills, prompts, commands, and hooks.
Use this decision model:
- Instructions: always-on repository or path-specific guidance.
- Prompt files and commands: reusable task prompts.
- Skills: portable workflows with optional scripts or resources, loaded on demand.
- Agents: persistent personas with tool, model, and permission boundaries.
- Hooks: deterministic lifecycle automation.
- MCP servers: external tools and data sources.
Do not create a new customization asset unless you can explain:
- why an existing asset is insufficient,
- which platform owns it,
- whether it should be canonical or generated,
- what other assets it depends on,
- how it will be validated.
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.
- 2d ago First seen · 38 lines · 30 tokens per session scan A 51fc08f78adb
agent-architect is an agent published in the GitHub repository configflux/weld (6 stars, last pushed 2d ago), licensed Apache-2.0. It adds 30 tokens to every session and 244 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 agents, from other repositories
architecture-review
Performs architectural review of files undergoing large-scale updates. Use proactively when refactoring, restructuring modules, or reviewing code organization. Provides actionable suggestions for renaming, moving functions, and restructuring modules.
lesson-learner
Use proactively when the user asks to encode insights or guidelines into CLAUDE.md. Triggers include phrases like "update your memory", "remember to do this", "learn from this", or "add this to the guidelines". Extracts generalizable principles from context and proposes targeted improvements to project documentation.
test-coverage-analyzer
Analyzes test coverage for a change set by reading source and test files, identifying which functions/modules have tests, and reporting coverage gaps. Use when reviewing PRs, before completing features, or when asked to check test coverage.
triage-investigator
Investigates a single entry point candidate to determine if it is a true positive, dead code, or a false positive that Ariadne missed callers for. Returns a TriageEntryResult JSON.
backlog-navigator
Researches the backlog to find incomplete tasks, recently completed tasks, and suggests the most likely next task candidates based on dependencies, priorities, and completion patterns. Use when starting work sessions or deciding what to work on next.
plan-reviewer
Reviews a synthesized fix plan from a specific angle (info-architecture, simplicity, fundamentality, or language-coverage). Writes review to a file.