bcquality/docs/agent-consumption.md

How agents consume BCQuality

BCQuality is content — knowledge files and skills. It is consumed by agents supplied by a host or orchestrator. This document explains the end-to-end flow so that skill authors, orchestrator maintainers, and contributors share one mental model.

Documentation | Partner quick start | Runner contract

This is the operational reference for integration authors. Partners using the installed plugin do not need to implement this flow themselves.

Try a minimal integration

"Invoke skills/entry.md" means ask your agent to read and follow that instruction document. It is not a shell command, HTTP endpoint, or executable library. Your host must be able to read files, enumerate directories, and execute the selected skills as instructed. Merely mentioning BCQuality does not make its content available to the model.

For a first integration, create or reuse a dedicated BCQuality checkout. For example, in PowerShell:

git clone https://github.com/microsoft/BCQuality.git "C:\Knowledge\BCQuality"

Give the host access to both that content directory and your own app directory. The plugin is not required for this route. Replace the paths and BC version below with your actual values, then send this prompt to the agent:

BCQuality root: C:\Knowledge\BCQuality
Review input: folder-path = C:\Repos\MyBusinessCentralApp

Read BCQuality's skills\entry.md and follow it with this task context:
task-context:
  goal: Review the complete AL app without changing its source files.
  inputs-available: [folder-path]
  technologies: [al]
  bc-version: 28
  enabled-layers: [microsoft, community, custom]
  disabled-skills: []

Resolve BCQuality instructions, knowledge, and index preparation against the
BCQuality root, not the app directory. Pass the actual review-input path above
when a dispatched skill accepts folder-path.
Follow Entry's preparation and dispatch instructions. Execute every dispatched
action skill with its exact input subset, reading READ and DO on demand.
Return each complete findings report unchanged. If Entry returns no-match or
failed, return that dispatch record unchanged instead of inventing a review.

inputs-available lists input types; the Review input line binds the type to the actual app directory. It is not an extra Entry schema field. Omit bc-version when unknown rather than guessing it; add localization or application-area context only when known. Keep the two roots distinct so index preparation operates on BCQuality, not your app.

Expect Entry to select the action skills and the agent to execute them. A broad review normally returns the Microsoft coordinator's report with domain sub-results, plus any separately dispatched reports. Each report must retain its outcome, including incomplete or failed work; see reading results. A dispatch record alone is not a completed review.

This prompt delegates the existing protocol rather than implementing new routing logic. For repeatable runs, pin the checkout. Add scheduling, retries, and rendering only when needed, using the runner contract.

The actors

• Orchestrator — the tool that triggers work. Lives outside BCQuality. Knows when to run something, not what to run.
• Agent — an LLM-driven process supplied by the host. It brings its own coding knowledge and tools; BCQuality adds curated guidance and execution contracts.
• BCQuality repo — two kinds of content:
  • Global skills in /skills/ — the entry.md entry-point skill plus the READ · DO · WRITE contracts that govern the rest of the repo.
  • Layer content in /microsoft/, /community/, and /custom/ — knowledge files and action skills grouped by authority.

When BCQuality is installed as a standalone plugin, it additionally exposes skills/al-code-review/SKILL.md and skills/al-development-plan/SKILL.md. These are host-format adapters, not additional action skills: each creates the task context and enters the same flow at Entry.

Repository structure

Path	Purpose
skills/entry.md	Routes a task to action skills.
skills/read.md, skills/do.md, skills/write.md	Stable knowledge, action-skill, and authoring contracts.
skills/al-code-review/SKILL.md	Host-format plugin adapter.
<layer>/knowledge/<domain>/	Atomic articles and optional sibling samples.
<layer>/skills/	Layer-owned action skills.
docs/	Partner guides and integration references.
evaluation/	Neutral review fixtures and scoring contract.
tools/	Knowledge-index and evaluation tooling.
.github/	Validation and repository workflows.

Layers are microsoft, community, and custom; Custom is a template for consumer forks. An action skill either evaluates knowledge directly (a leaf) or composes declared leaves (a super-skill). See global skills for the distinction between host-native packaging and these internal formats.

The flow

flowchart LR
    O[Host or orchestrator] -->|1 trigger + task context| A[Agent]
    A -->|2 invoke entry.md| E[Entry<br/>routing skill]
    E -->|3 dispatch record| A
    A -->|4 invoke dispatched skill| S[Action skill<br/>e.g. al-code-review]
    S -->|5 execute| P[Source → Relevance<br/>→ Worklist → Action<br/>reading READ · DO on demand]
    P -->|6 emit| R[Findings report<br/>or read-only guidance report]
    R -->|7 integrate| O

1. Orchestrator triggers

The orchestrator has a URL setting that points at BCQuality (default: github.com/microsoft/BCQuality) and a task to perform. It hands the agent a task context — goal, inputs available (pr-diff, file-path, …), technologies, BC version, enabled layers — and says: your source of truth lives at that URL; start by invoking /skills/entry.md.

2. Agent invokes Entry

The agent reads /skills/entry.md and runs it against the task context. Entry applies its Source → Relevance → Worklist → Action steps over the action skills under */skills/**/*.md and returns a dispatch record: the set of action skills to invoke, plus a list of candidates it skipped (with reasons). Routing is a skill, not orchestrator logic.

For a standalone plugin installation, the host activates the matching adapter first. The adapter preserves the caller's actual goal, constructs the task context, and invokes Entry. It does not select the internal review or plan-enrichment action skill itself or duplicate Entry's preparation, routing, and failure semantics.

3. Agent consumes the dispatch record

The dispatch record names one or more action skills, the subset of inputs each should receive, and each skill's output kind. The output kind distinguishes findings from read-only plan guidance before invocation; it is not proof of runtime side effects. If the outcome is no-match or failed, the agent returns the record to the orchestrator unchanged.

4. Agent invokes each dispatched action skill

Action skills live inside the layers — /microsoft/skills/, /community/skills/, /custom/skills/ — so their authority is carried by their location. For a PR review, Entry typically dispatches microsoft/skills/review/al-code-review.md. The agent reads the file and executes it.

5. Action skill executes the four-step pattern

Each action skill is a markdown file that specifies what to do at each step. The template is always the same:

Step	What happens
Source	Declare which knowledge folders and tags to search.
Relevance	Filter by frontmatter — bc-version, technologies, countries, application-area.
Worklist	Narrow from N candidates to the M that apply to this specific task.
Action	Apply the relevant knowledge and produce structured output.

Example: the Microsoft-owned performance review skill selects performance entries across every enabled layer, filters to bc-version: 26 and technologies: [al], narrows the candidate files to those that apply to the changed objects, and then evaluates each file against the diff. Its canonical corpus lives beside it under /microsoft/knowledge/performance/; cross-layer entries are limited to custom overrides or short-lived promotion work.

At this point the agent reads READ and DO on demand — it needs READ to interpret each knowledge file's frontmatter and sections, and DO to shape its output. Those contracts are fetched when first needed, not as part of bootstrap.

5a. The knowledge index (Source acceleration)

Discovering candidates at the Source step naively means opening every file under a domain folder just to read its frontmatter keywords — on a large corpus that is hundreds of file reads per review. To avoid this, BCQuality maintains a knowledge index: a single artifact (knowledge-index.json) that lists every article surviving the consumer's layer/allow-deny filtering and carries, per article, the exact inputs the Source/Worklist steps consume — path, layer, domain, frontmatter dimensions, keywords, title, and a one-line description hint.

The index is owned and produced by BCQuality, not reimplemented by each consumer. Its generator, tools/Build-KnowledgeIndex.ps1, ships here alongside the content. Entry ensures the index reflects the live tree before routing and regenerates it when absent or not known to be current. BCQuality CI validates that the generator is healthy and deterministic.

Consumers with allow/deny policy must prune their content copy before Entry runs. Building over that pruned tree prevents removed articles from entering discovery. A standalone plugin normally ships the whole tree: enabled-layers filters discovery but does not remove files or enforce a security boundary. See layer selection.

The index changes only how candidates are discovered, never which are selected. The Worklist predicate is unchanged — keywords still drive selection — and the agent still opens each worklisted article in full to read its ## Best Practice / ## Anti Pattern rule bodies; the index is discovery metadata only and never substitutes for the article body. When no index is present, skills fall back to path-based discovery (collect by domain folder), so review still works.

6. Agent emits structured output

The output contracts are defined in the DO meta-skill:

• A findings report carries review findings, domain labels, references, confidence, and suppressions.
• A development guidance report carries read-only knowledge constraints and validation considerations for an existing plan.

The orchestrator parses this without skill-specific logic. This is the point of the contract: orchestrators and action skills evolve independently.

For plan enrichment, the skill reads the existing plan and target repository, selects applicable knowledge, and returns constraints without changing the target. It does not generate a replacement plan, run tests, implement code, or drive a review/fix loop. Implementation stays in the consuming workflow.

7. Orchestrator integrates

The orchestrator turns findings into PR comments, build gates, or IDE diagnostics. It can feed read-only guidance into its own implementation phases, preserving all existing approvals and delivery gates.

Repository-specific development orchestrators

A repository-specific workflow can consume this read-only foundation before authoring while retaining its independent final review. This is the intended integration boundary, not a shipped consumer integration:

1. Investigate and produce the consumer's normal initial plan. Normalize any consumer-specific format outside BCQuality. A full serialized plan document containing metadata plus a markdown body (root cause or design intent, proposed changes, affected files, test strategy, acceptance criteria) is a valid boundary. A continuation/checkpoint payload is not a substitute for initial-plan coverage; workflow identifiers and state stay with the consumer.
2. Resolve and record an immutable BCQuality checkout and filtering policy. Invoke Entry with a read-only enrichment goal, the existing development-plan, repository, and established applicability dimensions. Keep index, guidance, and runner artifacts outside the target repository.
3. Execute the dispatched al-development-plan skill. Persist the unchanged report and provenance after any consumer state initialization or cleanup that could erase them. BCQuality does not own the state directory or lifecycle.
4. Inject relevant constraints and validation considerations into the existing Baseline, Implement, propagation (such as MiApp), and Critique phases, or equivalents. Re-enrich on material plan or applicability changes; preserve the relationship between plan version, guidance, and implementation attempt.
5. Run an independent final BCQuality review against the completed diff using the same recorded immutable checkout used for enrichment. Review the actual changes, not the guidance report as proof of correctness, then apply the consumer's ordinary delivery gates.

The consumer owns analysis, normalization, persistence, per-phase injection, approvals, TDD and runtime execution, propagation, retries, commits, and PR delivery. BCQuality supplies additional referenced product knowledge, not a replacement orchestrator.

Outcomes are additive, not a universal coding gate

no-knowledge with empty knowledge means no additional applicable BCQuality constraints. The consumer may proceed under its ordinary gates. It must not be conflated with failed retrieval/reference integrity (failed), incomplete evaluation or materially unresolved conditional guidance (partial), or absent required inputs (not-applicable). Consumers own the policy for handling those outcomes and recorded unknowns: seek missing context, re-enrich, escalate, or apply their existing risk controls without relabeling the report as successful. Do not fill gaps with generic articles simply to unlock implementation.

Pinning and pilot evidence

A configured tag or ref alone does not establish runtime pinning. Record the resolved commit and actual checkout/content identity used at invocation, along with enabled layers, pruning policy, index identity, plan version, and runner provenance. Verify that identity at both enrichment and final review; fetching default HEAD into an explicitly supplied checkout can bypass a configured ref. Use an isolated checkout that cannot drift during the run.

Consumer rollout and an external pilot remain follow-up work. A pilot must compare a pinned independent baseline run of the existing workflow without enrichment against a matched enriched run, keeping starting code, task, model, tools, runtime, and gates controlled and recording the actual BCQuality checkout. Retain external logs, diffs, test/compile outcomes, and independent final reviews, including failures and unresolved results. Credential-free fixture preparation and scorer regressions do not demonstrate improved repair quality, compilation, runtime success, or production integration.

Knowledge-backed and agent findings

BCQuality is an additive knowledge layer. The agent surfaces two kinds of findings, both shaped to the same DO output contract:

• Knowledge-backed findings carry one or more entries in references[] pointing at BCQuality knowledge files. Their id is the primary file's repo-relative path. Leaf sub-skills set domain to their human-readable display label, and super-skills preserve it verbatim during rollup.
• Agent findings carry an empty references: [], use a slug id prefixed agent:, and have confidence capped at medium. A leaf can emit one strictly within its own domain and uses that leaf's display label. A super-skill can emit a cross-cutting agent finding with from-sub-skill: "agent" and domain: "Agent". Their message is self-contained because there is no knowledge-file footer to fall back on.

Before a skill emits an agent finding, it validates the candidate against the BCQuality knowledge already loaded for the task: a matching file upgrades the candidate to a knowledge-backed finding (and merges or deduplicates against relevant existing output); a contradicting file suppresses the candidate. Only candidates with no BCQuality coverage become agent findings.

Orchestrators MUST tolerate an absent domain in reports from older producers. When it is present, treat it as display text rather than an identifier: preserve the full string and its case, whitespace, punctuation, and non-ASCII characters, escaping only for the target rendering format. Do not tokenize it on spaces or use a lowercased or slugified form as the sole metadata or deduplication key, because distinct labels can collapse to the same slug. Retain the exact string, use a lossless encoding, or use a collision-resistant digest instead. Orchestrators MAY render knowledge-backed and agent findings differently and MAY apply independent severity floors; references: [] and the agent: id prefix distinguish agent findings, while from-sub-skill: "agent" identifies those emitted by the super-skill itself.

Why this architecture

• Entry is the only hardcoded thing. Orchestrators ship with one convention — "invoke /skills/entry.md first" — and nothing else. New action skills and new knowledge files are picked up automatically because Entry discovers them at dispatch time.
• Standalone installation adds an adapter, not another policy layer. The plugin's host-format al-code-review skill only translates the invocation into Entry's task context. Entry and the dispatched action skills remain authoritative.
• Layers decide authority, not code. The agent sees /microsoft/ and /community/ together; if two files conflict, the precedence rule defined in READ resolves it. A partner fork can disable /community/ — that's a config choice, not a code change.
• Knowledge and skills evolve independently within their owning layer. A new knowledge file requires no skill changes because existing skills pick it up via frontmatter filters. Layer placement still follows skill ownership, so promoting a skill also promotes its canonical corpus.

The mental model, in one sentence

The orchestrator knows when to run; Entry decides which skill to run; the action skills define what to do; the meta-skills teach the agent how to behave; the knowledge files are what the agent knows.