AgentStack
SKILL verified Apache-2.0 Self-run

Trd Gen

skill-neplich-dev-agent-skills-trd-gen · by Neplich

Internal engineering specialist—not a direct entry point. Invoked by engineer-agent after pm-agent handoff to create or update Engineer-owned TRDs, API docs, ADRs, and implementation blueprints from stable PM decisions.

No reviews yet
0 installs
11 views
0.0% view→install

Install

$ agentstack add skill-neplich-dev-agent-skills-trd-gen

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Trd Gen? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

TRD Generator

Engineer-owned technical planning skill. It turns confirmed PM requirements into docs/engineer/{feature_path}/TRD.md, mirroring docs/pm/{feature_path}/PRD.md. When the confirmed technical scope includes interface contracts or architecture decisions, it also owns docs/engineer/{feature_path}/API.md and docs/engineer/{feature_path}/ADR--.md, then hands the confirmed Engineer document set to feature-implementor for an implementation plan and code execution.

Role Boundary

trd-gen owns:

  • technical approach and architecture trade-offs
  • module and file impact analysis
  • interface, data, deployment, observability, and validation strategy
  • engineering risks, blockers, assumptions, and open technical questions
  • writing or updating docs/engineer/{feature_path}/TRD.md
  • writing or updating docs/engineer/{feature_path}/API.md when interface

contracts are in scope

  • writing or updating docs/engineer/{feature_path}/ADR-*.md when durable

technical decisions are in scope

  • resolving TRD gap packets from discoverers such as engineer-agent,

debugger, or feature-implementor

trd-gen does not own:

  • PM scope, user stories, business acceptance criteria, or product decisions
  • UI/UX or visual design decisions
  • code implementation
  • implementation plan documents produced after TRD approval

If the PRD, product decisions, or acceptance scope is not stable, stop and hand back to pm-agent:idea-to-spec with the missing decisions. DECISIONS.md is a valid decision record when present, but equivalent confirmed product decisions are also acceptable.

If the target agent's plugin for a cross-agent handoff is not installed or unavailable, state the missing stage and required plugin, mark that handoff stage as blocked, and do not perform the missing agent's responsibilities yourself.

When another skill hands back a missing, incomplete, stale, or conflicting TRD, the discoverer owns describing the TRD gaps and trd-gen owns completing the TRD. Treat the handoff as a gap packet, not as an implementation request.

PM Handoff Entry Gate

Before writing or updating Engineer-owned documents, require a PM handoff packet or equivalent confirmed PM documents with stable product scope and feature path. If the user directly invokes trd-gen with an unresolved product idea, expectation change, or ambiguous feature path, stop and return the request to pm-agent for classification. Direct invocation does not bypass the need for stable PM scope.

Use the PM-side packet definition in agents/product_manager/skills/idea-to-spec/_internal/_shared/skill-map.md.

Required Flow

flowchart LR
    PM["pm-agent: PRD / BRD / product decisions confirmed"] --> Handoff["Explicit handoff to engineer-agent:trd-gen"]
    Handoff --> PathGate["Resolve feature_path from docs/pm/{feature_path}/PRD.md"]
    PathGate --> TRD["trd-gen writes docs/engineer/{feature_path}/TRD.md"]
    TRD --> OptionalDocs["API.md / ADR-*.md when in scope"]
    OptionalDocs --> Review["Maintainer confirms Engineer docs"]
    Review --> Plan["Explicit handoff to feature-implementor"]
    Plan --> ImplPlan["feature-implementor writes docs/engineer/{feature_path}/IMPLEMENTATION_PLAN.md"]
    ImplPlan --> Code["Implementation / tests / delivery"]

Use this checkpoint language:

PRD 已确认,当前进入 Engineer TRD 阶段。
我会基于 PRD、产品决策记录和仓库上下文解析 `feature_path`,并编写
`docs/engineer/{feature_path}/TRD.md`。
Engineer 文档确认后,再移交给 `feature-implementor` 编写实现计划文档并进入实现。

TRD Gap Packet Handling

Accept a TRD gap packet from engineer-agent, feature-implementor, debugger, or QA E2E alignment when PM scope is stable but the TRD is missing, incomplete, stale, or conflicts with implementation or test evidence.

The incoming packet should identify:

  • source request, feature, and PRD / decision records already checked
  • affected components, modules, APIs, data flow, integrations, or deployment

surfaces

  • missing or conflicting technical decisions
  • validation commands or evidence that exposed the gap
  • release, rollback, observability, security, error-handling, or E2E coverage

risk when relevant

  • the discoverer's boundary statement: the finder names the gaps; trd-gen

completes or updates the TRD

trd-gen must either update docs/engineer/{feature_path}/TRD.md to resolve each named gap or record an open technical question with owner, blocker, and unblock condition. Do not route to feature-implementor, debugger, or QA E2E documentation updates until the TRD is confirmed, mirrors the PRD feature path, and any open questions are explicitly accepted as non-blocking.

Document-Writing Delegation

To avoid context drift during long document drafting, all TRD writing and TRD revision work must be delegated to a fresh document-writing sub-agent when sub-agent capabilities are available.

The main process keeps the source context and final judgment. The delegated document-writing task must include:

  • PRD, BRD, DECISIONS.md when present, equivalent product decisions, design

docs, and relevant issue links

  • current codebase and repository constraints
  • any TRD gap packet from the finder, including affected components, data flow,

validation, release risk, and error-handling gaps

  • required output path: docs/engineer/{feature_path}/TRD.md
  • optional Engineer-owned output paths:

docs/engineer/{feature_path}/API.md and docs/engineer/{feature_path}/ADR-*.md

  • forbidden areas and instruction not to implement code
  • required output: changed document path, summary, assumptions, open questions,

and validation notes

After the sub-agent returns, the main process reviews the TRD for requirement traceability, technical completeness, repository fit, and unresolved blockers before asking for TRD confirmation.

Inputs

  • Required:
  • confirmed PRD or equivalent approved requirement document
  • DECISIONS.md or confirmed product decisions
  • repo path and current system context
  • resolved feature_path, parent_feature, and feature_level from the PRD

or PM handoff

  • Optional:
  • BRD
  • design specs
  • existing API / ADR / deployment docs
  • issue or PR references
  • preferred stack or explicit technical constraints

Output

Write or update:

docs/engineer/{feature_path}/TRD.md
docs/engineer/{feature_path}/API.md                 # when API docs are in scope
docs/engineer/{feature_path}/ADR--.md     # when an ADR is in scope

The TRD must include:

  • metadata with type: TRD, feature, feature_path, parent_feature,

feature_level, version, date, last_updated, and related_prd

  • source documents and requirement traceability
  • technical overview and architecture diagram
  • impacted modules, components, APIs, data, and integration points
  • API documentation when interface contracts are stable enough to document
  • ADRs when a technical decision needs durable rationale
  • implementation constraints and non-goals
  • validation strategy and concrete verification commands when known
  • rollout, observability, security, and operational concerns when applicable
  • risks, assumptions, and open technical questions
  • explicit handoff conditions for feature-implementor

When updating a TRD from a gap packet, address each named gap directly or record it as an open technical question with the owner and unblock condition.

Quality Checks

Before handoff, verify:

  1. Every P0 PRD requirement maps to a technical component or explicit non-goal.
  2. Technical decisions do not change PM scope.
  3. Unknowns are marked as assumptions or open questions, not hidden as facts.
  4. The TRD path is under docs/engineer/{feature_path}/ and mirrors

docs/pm/{feature_path}/PRD.md.

  1. feature_path, parent_feature, and feature_level match the PRD. Old

single-level PRDs without these fields may be read as feature_path=, parent_feature=N/A, and feature_level=1.

  1. related_prd points to docs/pm/{feature_path}/PRD.md.
  2. API and ADR documents, when produced, live under

docs/engineer/{feature_path}/ and do not use only the terminal feature name as a parallel top-level directory.

  1. Any inbound TRD gap packet has been resolved or explicitly tracked as open.
  2. The next step is feature-implementor only after the Engineer document set

is confirmed.

Handoff

After the TRD is confirmed:

Engineer 文档已确认,当前移交给 `feature-implementor`。
下一步应基于 `docs/engineer/{feature_path}/TRD.md` 编写
`docs/engineer/{feature_path}/IMPLEMENTATION_PLAN.md`,确认后再进入代码实现。

Do not continue into implementation unless the user explicitly confirms the TRD or asks to proceed despite open technical questions.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.