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 skills add rubenzarroca/sdd-plugin --skill sdd-implementgit clone --depth 1 https://github.com/rubenzarroca/sdd-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/skills/rubenzarroca/sdd-plugin/sdd-implement)<a href="https://agentmods.dev/skills/rubenzarroca/sdd-plugin/sdd-implement"><img src="https://agentmods.dev/badge/skills/rubenzarroca/sdd-plugin/sdd-implement/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/skills/rubenzarroca/sdd-plugin/sdd-implement"><img src="https://agentmods.dev/badge/skills/rubenzarroca/sdd-plugin/sdd-implement.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 | $0.00073 | $0.04062 |
| Opus 5.5 | $0.00029 | $0.01625 |
| Sonnet 5.5 | $0.00015 | $0.00812 |
| Haiku 4.5 | $0.00007 | $0.00406 |
Grade A, and why
sdd-implement 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 — 346 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/sdd:implement — Execute tasks
You are implementing tasks from the SDD task list. This is the most tightly scoped command in the plugin. You read only task blocks and the files listed in them. In single-task mode, execute one task. In batch mode, execute multiple small tasks in a single pass. Follow these steps exactly, in order. Do NOT skip steps. Do NOT read files beyond what is explicitly listed in the tasks.
MCP Server integration: This command uses the sdd-server MCP tools for state queries and feature transitions. If the MCP server is not available, fall back to reading/writing .sdd/state.json directly.
Coaching Layer
During implementation, the user shifts from active participant to observer. They watch code being written but may not understand what is being created or why. Without coaching, this phase becomes a black box. Claude must make the process transparent.
Rules:
- Announce before you build. Before starting implementation, briefly tell the user what you're about to do in plain language: "I'm going to create [file] which handles [function in business terms]. This implements [FR-xxx] from your spec — [what that requirement means in their domain]."
- Connect back to the spec. In the post-task report, always reference which requirements were addressed: "This task implemented FR-003 (authentication on all protected routes) and EC-002 (invalid token handling)." The user should never wonder "why did that just happen?"
- Explain blockers in user terms. If a blocker is found, explain it without jargon: "I found something this task needs that isn't part of its instructions. [Plain description]. We have two options: add it to an existing task, or create a new one. What would you prefer?"
- Explain the single-feature lock if triggered: "Only one feature can be implemented at a time to prevent conflicting changes. Feature
{other}is currently being built. Complete or pause it before starting this one." - Calibrate explanation depth. Read
completed_featuresfromsdd_get_state:0(first feature): Detailed plain-language explanations of what each file does and why.1(second feature): Brief explanations — focus only on what's novel or surprising about this task.2+: Minimal — announce what you're building in one line, report results. The user knows the process; don't slow them down with explanations they don't need.
- Role-transition coaching. Check
.sdd/state.jsonfieldmilestones.role_transition_explained. Iffalseand this is the first task for this feature, explain the role shift: "We're starting implementation now. From here, I'll work through each task one at a time and report what I built. Your role shifts to reviewer — check that what I built matches what you described in the spec. If anything looks wrong or confusing, stop me and ask." Then setmilestones.role_transition_explainedtotrue. On subsequent features, skip — the user knows the drill.
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 First seen · 346 lines · 73 tokens per session scan A f7a0f05b966c
sdd-implement is a skill published in the GitHub repository rubenzarroca/sdd-plugin (7 stars, last pushed 7mo ago), licensed MIT. It adds 73 tokens to every session and 4,062 once invoked, about $0.0003 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 skills, from other repositories
review
Business review — weekly health check, monthly close, or ad-hoc audit. Subcommands: weekly, monthly, check. Use for periodic reviews or metrics assessment.
scheduler
Manage scheduled jobs — create, update, list, run remote agents on cron. Use when setting up recurring tasks, checking job status, or managing automation.
do
Alias for /brana:backlog start with freeform text. Routes to the best skill or creates a task. Use /brana:backlog start directly for the same behavior.
github-issue-triage
A GitHub issue organiser for open problems and requests in a software project. GitHub is a service where teams store code and track issues.
backlog
Manage the backlog — plan, track, navigate phases and epics. Use when planning phases, viewing roadmaps, or restructuring work.
onboard
Scan and diagnose a project, or scaffold a new client from scratch. Works for code and venture clients. Auto-detects project type.