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 swan-gtm/gtm-skills --skill leaked-demand-recoverygit clone --depth 1 https://github.com/swan-gtm/gtm-skillsWrote 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/swan-gtm/gtm-skills/leaked-demand-recovery)<a href="https://agentmods.dev/skills/swan-gtm/gtm-skills/leaked-demand-recovery"><img src="https://agentmods.dev/badge/skills/swan-gtm/gtm-skills/leaked-demand-recovery/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/swan-gtm/gtm-skills/leaked-demand-recovery"><img src="https://agentmods.dev/badge/skills/swan-gtm/gtm-skills/leaked-demand-recovery.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00155 | $0.01252 |
| Opus 5 | $0.00077 | $0.00626 |
| Sonnet 5 | $0.00031 | $0.00250 |
| Haiku 4.5 | $0.00015 | $0.00125 |
Grade A, and why
leaked-demand-recovery 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 9d 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 — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The cheapest pipeline in any organisation is demand already paid for and then lost in the gap between two systems. This finds it, proves the drop is real before anyone touches it, and drafts the recovery.
The play
-
Scope the sweep. Choose the seams to work and the window: about a month for active leaks, six to twelve months for closed-lost revival. Confirm the window if one was specified.
-
Build the candidate list from the capture systems, not the CRM. The CRM holds what someone remembered to log. The mailbox, the calendar, the call recorder and the lead-capture channel hold what actually happened. Take the union, deduplicate by person and by company, and include people with no CRM record at all: they are the largest silent category and the entire reason to build the list this way.
references/leak-map.mdlists the seams worth working and what each is verified against. -
Verify every candidate against two independent systems. This is a hard gate. The record saying someone was dropped is frequently wrong, and the cost is asymmetric: a missed recovery forfeits one opportunity, while messaging someone who already booked or already replied costs the relationship. Read
references/verification-gate.mdbefore running this step. Candidates that fail the gate leave the list with a one-line note. Never soften a failed gate into a light touch anyway. -
Establish history before choosing a voice. A recovered hand-raiser is usually not new. No history: fresh first touch. Prior conversation: re-engagement voice that references the context honestly. Existing or churned customer: route to the account owner and stop drafting. Open deal owned by a colleague: flag the owner and stop, because a parallel message collides with a live motion. Closed lost: carry the stated reason into the framing, and proceed only if the blocker has genuinely changed.
-
Apply hard disqualifiers before scoring fit. Geography, regulatory coverage, segment. Serviceability is binary and cheap; fit scoring is expensive and pointless on an account that cannot be served. Verify thin enrichment rather than trusting it, since poor records under-report company size. Keep the disqualifier list in one place, or a stale gate rejects markets since opened and admits ones since dropped.
-
Draft one message per confirmed drop, personalised on what they asked about or the exact point the last conversation reached. The recovery framing does the work; generic re-engagement copy wastes the advantage of knowing precisely where they fell. Where the drop was an internal failure, acknowledge it in a clause and move on. One unambiguous next step.
-
Present for approval. Per confirmed drop: person and company, leak type, evidence checked, history, suggested action. Then the drafts. Keep failed-gate candidates visible at one line each so the verification work is auditable and a bad call can be spotted. On a scheduled sweep that finds nothing, end silently.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 9d ago First seen · 89 lines · 155 tokens per session scan A aba21b27ac7f
leaked-demand-recovery is a skill published in the GitHub repository swan-gtm/gtm-skills (150 stars, last pushed 2d ago), licensed MIT. It adds 155 tokens to every session and 1,252 once invoked, about $0.0008 per session on Opus 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 2026-09-03.
Other skills, from other repositories
leadmagic-cli
Use the LeadMagic CLI for enrichment automation — find emails, validate contacts, enrich CSVs in batch, find decision-makers by role, and push to outbound platforms (Smartlead, Instantly). Use when automating enrichment from the command line, batch-processing contacts, or integrating LeadMagic into scripts. Triggers…
revenue-team-onboarding
Revenue team onboarding — security access provisioning, ramp schedules, 30-60-90 plans, Slack engagement, shadowing, certification, and time-to-first-meeting reduction. Use when onboarding SDRs, AEs, managers, or RevOps hires. Triggers on: "sales onboarding", "SDR ramp", "30-60-90 sales", "reduce ramp time", "Slack…
team-management
Team management for startup leaders — 1:1s, performance reviews, goal setting (OKRs), delegation, feedback culture, remote team management, team rituals, and scaling from IC to manager. Use when building team processes, managing first direct reports, or scaling management as the company grows. Triggers on: "team…
brief-me-for-today
Wake up with a one-screen brief: today's meetings, drafts waiting on your sign-off, the top three moves I'd make today, and your watch list of stalled deals or red customers. Calendar is the anchor; I never invent meetings.
Follow-ups
Manage follow-up sequences.
meeting-type-management
Manages Chili Piper team meeting types and their email/SMS reminders — list, inspect, create, update, delete — with dry-run planning, guest-visible-field safety (inviteTitle/inviteDescription vs internal description), and reminder attach/detach.