Frequently Asked Questions

May 14, 2026 · View on GitHub

General

What is this repo?

A collection of ready-to-use GitHub Agentic Workflows for non-engineering roles — PMs, compliance teams, and leaders. The workflows read your existing GitHub Issues and Projects to generate readiness reports, compliance reviews, and go-to-market content automatically.

Do I need to write code to use these workflows?

No. You customize behavior by editing markdown policy files in .github/policies/. The workflows themselves are compiled from markdown using gh aw compile — no YAML authoring required.

How is this different from The Agentics?

The Agentics focuses on engineering workflows (CI, code review, testing). Agentics Beyond Code covers the non-engineering side — launch tracking, compliance, and go-to-market content.


Setup & Configuration

What do I need to get started?

A GitHub repository with Issues and a Project board. See the Getting Started guide for prerequisites, token setup, and first-run instructions.

Can I use this with an existing project?

Yes. The workflows read from a GitHub Project and its linked issues. As long as your issues use the expected labels (launch, initiative, epic) and your project has the required custom fields (Phase, Target Date), everything works. See setup.md for the full list.

How do I change when a compliance review is required?

Edit the rubric section of the relevant policy file in .github/policies/ (e.g., security-review-policy.md). Changes take effect on the next workflow run — no recompilation needed.

How do I change the writing style for GTM content?

Edit .github/policies/voice-and-tone-policy.md. The GTM Content workflow reads it at runtime and applies it to all generated drafts.


Workflows

How often do the workflows run?

WorkflowScheduleManual trigger?
Launch Readiness CheckerMonday ~8:30 AM PTYes
Compliance ReviewMonday ~8 AM PT and relevant issue labelsYes
Compliance Team ReportsMonday ~8:45 AM PTYes
GTM ContentMonday ~8:15 AM PTYes
GTM Team ReportsMonday ~8 AM PTYes
Leadership BriefsMonday ~7:30 AM PTYes
Assumption SurfacerOn issue opened/editedYes
Intake Request TriageOn issue labeled triage-neededNo
Decision LogWeeknights after each workdayYes
Daily Standup PrepMonday and Wednesday ~8 AM PTYes
Strategy AlignmentWednesday ~8 AM PTYes
Adversarial PMWednesday ~8:30 AM PTYes
Weekly StatusFriday ~8 AM PTYes
Workflow HealthFriday ~8 AM PTYes
Transcript ProcessorOn push to /transcripts/Yes
Process AnalyzerMonday night / early Tuesday UTCYes
Sample Data SimulatorSunday and Tuesday nightsYes

The Sample Data Simulator is for demo purposes only — it generates fake project activity so the other workflows have realistic data. You don't need it for production use. Trigger it manually anytime with gh aw run sample-data-simulator to add more content.

All workflows can also be triggered on demand with gh aw run <workflow-name>.

What happens if I run a workflow twice?

Workflows are idempotent. They check for existing sub-issues, labels, and comments before creating new ones. On a second run, existing content is updated in place rather than duplicated.

Can workflows conflict with each other?

Yes — workflows that run on the same schedule or modify the same resources (issues, labels, discussions) can interfere. For example, Monday-morning workflows may read stale data if another workflow is still updating labels. The Workflow Health report includes a Cross-Workflow Interactions section that detects these conflicts automatically each week: concurrent runs, shared resource modifications, label churn, and cascade chains (one workflow's output triggering another). Check the Friday health report for any 🔴 high-risk interactions and follow the recommended mitigations.

Will workflows create issues I don't want?

Workflows only create sub-issues when their rubric or policy says one is needed. You control the criteria by editing the policy files. Sub-issues are always created as children of the relevant launch — they won't clutter your repo with orphaned issues.

Do generated sub-issues get published automatically?

No. Compliance review sub-issues and GTM drafts are created as open issues for the DRI or reviewer to act on. Changelog drafts and roadmap items are explicitly marked as drafts that need human review before publishing.


Compliance

Which compliance teams are supported?

Four teams out of the box: Security, Privacy, Accessibility, and Responsible AI. Each has its own policy file with a rubric, review questions, and checklist.

Can I add or remove compliance teams?

Yes, but it requires editing the workflow file (compliance-review.md) and creating or removing the corresponding policy file in .github/policies/. After changes, recompile with gh aw compile compliance-review.

How do compliance labels work?

  • ai:needs:{team} (e.g., ai:needs:security) — the workflow adds this when a review is required
  • approved:{team} (e.g., approved:security) — a reviewer adds this manually when the review passes
  • The Launch Readiness Checker factors these labels into its risk assessment

GTM Content

What content does the GTM workflow generate?

Two sub-issues per launch:

  • Changelog draft — a customer-facing announcement post (created for Alpha phase and later)
  • Roadmap item — a forward-looking description for public roadmaps (created for all phases)

Why does the roadmap item only show a quarter, not a specific date?

Roadmap items are customer-facing. Sharing a specific internal target date creates false precision and sets expectations the team may not be able to meet. A quarter (e.g., "Q3 2026") communicates timing without over-committing.


Decision Log

How does the decision log detect decisions?

The workflow scans issue comments and transcript files for decision-signal language — phrases like "we decided to…", "the call is…", "going with option…", or "final decision:". It uses context clues to extract the decision, options considered, rationale, and participants.

Where are decision records stored?

Individual markdown files in the /decisions/ directory, named YYYY-MM-DD-<slug>.md. The workflow creates a PR with the new files so you can review before merging.

Can I write decision records manually?

Yes. The /decisions/ directory is just markdown files. You can create them manually following the same format, or edit the ones the workflow generates.


Transcript Processing

What transcript formats are supported?

Plain text (.txt) and WebVTT (.vtt) files. Drop them into the /transcripts/ directory and push to trigger the workflow.

How does the transcript processor match to issues?

It extracts key topics, names, and action items from the transcript and fuzzy-matches them against open issue titles, labels, and body content. Only high-confidence matches get comments posted.

Will it create duplicate comments if I push the same transcript again?

The workflow checks for existing comments from the same transcript file to avoid duplicates.


Process Analyzer & Weekly Retro

What does the Process Analyzer do?

It runs weekly and reads all meeting transcripts from the past 7 days. It does three things:

  1. Posts a weekly retro Discussion focused on how the team is working (collaboration patterns, blockers with process root causes, team energy, process health) — not what shipped (the Weekly Status workflow covers that)
  2. Detects process drift by comparing what people describe in transcripts against docs/how-we-work.md. When drift is found, it creates a PR to update the doc
  3. Identifies automation candidates — manual processes people mention that could be handled by a workflow — and files issues for strong candidates

What is docs/how-we-work.md?

A living document describing how the team operates — meeting cadence, issue triage, PR review SLAs, on-call rotation, communication norms, and an inventory of what's already automated vs. still manual. The Process Analyzer keeps it current by detecting when the team's actual behavior drifts from what's documented.

⚠️ The included docs/how-we-work.md is sample data for demo purposes. Replace it with your actual team's processes before using in production.

How does it detect process drift?

It looks for discrepancies between transcript discussion and the how-we-work doc — for example, someone saying "we moved design review to Thursday" when the doc says Wednesday, or "we're doing rotating triage now" when the doc says one person handles it. Only decided changes trigger a PR update; casual suggestions are noted but don't modify the doc.

How is the weekly retro different from the Weekly Status report?

The Weekly Status covers what the team shipped, learned, and needs help with — it's a leadership status rollup. The Process Analyzer retro covers how the team is working — collaboration patterns, process friction, team energy, and whether documented processes match reality. They complement each other.


Strategy Alignment

What is docs/strategy.md?

A document listing the team's strategic tradeoffs as "X, even over Y" statements — both sides are valuable, but when they conflict, the team leans toward X. The Strategy Alignment workflow keeps it annotated with real evidence from the team's decisions and activity.

⚠️ The included docs/strategy.md is sample data for demo purposes. Replace it with your actual team's strategic tradeoffs before using in production.

How does the workflow decide what's misaligned?

It applies a high bar: the contradiction between a decision and a tradeoff must be stark and unambiguous. Borderline cases are skipped. If a decision acknowledges the deviation ("we know this goes against our usual approach but…"), the workflow treats it as intentional and doesn't comment.

Will it spam my issues with comments?

No. The workflow comments on a maximum of 2 issues per run, and only for clear misalignment. Most weeks it produces zero comments — just a PR annotating the strategy doc with alignment evidence.


Costs & Tokens

Do workflows cost money to run?

Workflows currently use OpenAI Codex via GitHub Actions. Model usage is billed to the OpenAI account associated with the OPENAI_API_KEY or CODEX_API_KEY repository secret. Each workflow run includes token usage in the Actions artifacts so you can track spend.

Can I control costs?

Yes. Workflows run on a weekly schedule (Monday/Friday mornings PT) — you can adjust the schedule or run them only on demand. Each workflow also has a timeout-minutes setting to cap execution time.