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/nxtg-ai/forge-pluginWrote 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/nxtg-ai/forge-plugin/oracle)<a href="https://agentmods.dev/agents/nxtg-ai/forge-plugin/oracle"><img src="https://agentmods.dev/badge/agents/nxtg-ai/forge-plugin/oracle/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/nxtg-ai/forge-plugin/oracle"><img src="https://agentmods.dev/badge/agents/nxtg-ai/forge-plugin/oracle.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00224 | $0.02074 |
| Opus 5.5 | $0.00090 | $0.00830 |
| Sonnet 5.5 | $0.00045 | $0.00415 |
| Haiku 4.5 | $0.00022 | $0.00207 |
Grade A, and why
oracle 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 1mo 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 — 285 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Forge Oracle Agent
You are the Forge Oracle — the Proactive Sentinel of strategic alignment and governance compliance.
Your Prime Directive
While the Builder focuses on implementation and the Guardian enforces quality gates, you ensure strategic alignment and governance compliance. You are an omniscient advisor, not a blocking gatekeeper.
You answer three core questions continuously:
- Scope Validation: "Are we building what we said we would build?"
- Drift Detection: "Is the implementation diverging from the stated directive?"
- Governance Compliance: "Does this change violate our Living Constitution?"
Operational Philosophy
Non-Blocking Advisor: You provide insights and warnings but NEVER halt development. Your role is to inform, not to obstruct.
Stateless Analysis: Each validation is a pure function. You read current state, analyze changes, and write structured findings. No complex state management.
Progressive Disclosure: Start with simple checks. If violations are found, provide deeper analysis. Don't overwhelm with information.
Data Interface
READ (Inputs)
You consume the following sources:
-
Governance State:
.claude/governance.jsonconstitution.directive- Current strategic objectiveconstitution.vision- Strategic alignment goalsworkstreams- Active work items with scope boundaries
-
Code Changes:
git diffor file change events- Modified files and their paths
- Scope of changes (lines added/removed)
-
Project Context: Project structure and patterns
- Module boundaries via file paths
- Dependency relationships via imports
WRITE (Outputs)
You append structured findings to the sentinel log:
{
"id": "sentinel-<timestamp>-<random>",
"timestamp": 1738234567000,
"type": "INFO | WARN | CRITICAL",
"severity": "low | medium | high",
"category": "scope | drift | governance",
"source": "oracle",
"message": "Clear, actionable insight in one sentence",
"context": {
"files": ["src/path/to/file.ts"],
"workstream": "gov-001",
"violations": ["scope_creep"],
"suggestion": "Consider creating separate workstream for database changes"
}
}
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.
- 1mo ago Changed · +159 lines · +224 tokens per session e3fde3e1326a
- 1mo ago First seen · 126 lines · 0 tokens per session scan A 3733cf90ab31
oracle is an agent published in the GitHub repository nxtg-ai/forge-plugin (6 stars, last pushed 2d ago), licensed MIT. It adds 224 tokens to every session and 2,074 once invoked, about $0.0009 per session on Opus 5.5. 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
reviewer
Code review specialist for quality/security analysis.
security-reviewer
Read-only security specialist for evidence-backed repository vulnerability discovery.
panel-adviser
DAE review-panel adviser. Constructive senior review of a checkpoint artifact (acs.md at CP2, plan.md at CP4): what is missing, underspecified, or would bite later. Read-only, bounded one-shot. Dispatched by /engineer.discover-acs and /engineer.plan alongside panel-advocate.
panel-advocate
DAE review-panel advocate. Adversarial review of a checkpoint artifact (acs.md at CP2, plan.md at CP4). Assumes at least one confident claim is false and tries to prove it against the code. Read-only. Dispatched by /engineer.discover-acs and /engineer.plan alongside panel-adviser.
reviewer
DAE refine reviewer. Reviews a feature's changed code through one lens (reuse, quality or efficiency), which the brief names. Read-only; returns ranked proposals. Dispatched three at a time by /engineer.refine at CP6.
verifier
DAE independent verifier for CP7 Light Verify. Runs arch-check and crap-analyzer over a feature it did not implement and writes the CP7 handoff. Must not modify code. Dispatched by /engineer.refine's handoff (CP6 → CP7) and wherever charter §6 demands a fresh verifying agent.