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 getaero-io/gtm-eng-skills --skill deepline-ads-audiencesgit clone --depth 1 https://github.com/getaero-io/gtm-eng-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/getaero-io/gtm-eng-skills/deepline-ads-audiences)<a href="https://agentmods.dev/skills/getaero-io/gtm-eng-skills/deepline-ads-audiences"><img src="https://agentmods.dev/badge/skills/getaero-io/gtm-eng-skills/deepline-ads-audiences/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/getaero-io/gtm-eng-skills/deepline-ads-audiences"><img src="https://agentmods.dev/badge/skills/getaero-io/gtm-eng-skills/deepline-ads-audiences.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 7 findings, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Memory Poisoning · line 99 Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.Fix: Implement context-window management that detects and rejects padding or stuffing attempts. Prioritize system instructions over user-injected content.
- medium Prompt Injection · line 99 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
- medium Prompt Injection · line 99 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
- medium Prompt Injection · line 123 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
- medium Prompt Injection · line 119 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
- medium Prompt Injection · line 119 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
- medium Prompt Injection · line 121 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
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 | $0.00133 | $0.08155 |
| Opus 5 | $0.00067 | $0.04078 |
| Sonnet 5 | $0.00027 | $0.01631 |
| Haiku 4.5 | $0.00013 | $0.00816 |
Grade A, and why
deepline-ads-audiences scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 498 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Deepline Ads Audiences
Quick Start
npm install -g deepline
# Fallback for secure sandboxes: mkdir -p "$HOME/.local" && npm config set prefix "$HOME/.local" && export PATH="$HOME/.local/bin:$PATH" && npm install -g deepline --registry https://code.deepline.com/api/v2/npm/
deepline auth register --wait auto
deepline auth wait --timeout 120 # completes Cowork/browser approval; no-op if already connected
deepline auth status
deepline -h
CLI resolution
Run deepline when it is available. If the shell reports that command is missing, use <workspace-root>/.deepline/runtime/bin/deepline (or the npm-created .cmd shim on Windows). If neither exists, follow https://code.deepline.com/INSTALL.md to set up Deepline.
Build high-quality ABM paid ads audiences from first-party customer or prospect lists. This skill is for paid ads audience upload and evaluation, not outbound.
Names in this skill are starting hints. Run deepline tools search audience --json and deepline tools describe <tool_id> --json before executing because tool names and payload shapes can change. Tool search accepts an intent query or, for structured filtering, --categories and/or --search_terms; a filter-only search needs at least one of those flags. Use commas for multiple filter values, and put provider names in the query rather than using a --prefix flag.
Before You Start
Use the full recipe when the user asks to enrich and upload audiences to Facebook/Meta and Google:
→ Read recipes/enrich-and-upload-facebook-google.md.
Use the max-coverage recipe when the user asks for "max coverage", "maximum match rate", "keep increasing coverage", "get to 75% coverage", or asks to exhaust LinkedIn/personal-email/hash options:
→ Read recipes/max-coverage-audience.md.
This skill is not for cold outbound, sequencing, or copywriting. Personal emails here are used to improve paid ads matching, not to contact people directly.
Decision Matrix
| User says | Do this | Read |
|---|---|---|
| "max coverage", "highest match rate", "keep increasing coverage" | Run the explicit max-coverage ladder with budget gates. | recipes/max-coverage-audience.md |
/deepline-ads-audience, "enrich and upload to FB/Google" |
Run the full paid ads audience recipe. | recipes/enrich-and-upload-facebook-google.md |
| "sample ABM segment", "do the example workflow" | Follow the reusable high-priority ABM segment recipe. | recipes/sample-abm-segment-example.md |
| "use ContactOut hashes", "hashed identifiers", "LinkedIn URLs to hashes" | Plan a bulk pass beside the ladder, not a waterfall step. | shared/contactout-hash-pool.md |
| "what is a hash", "why is my match rate low", first-time user | Explain the mechanic before quoting a plan. | shared/audience-basics.md |
encoded/internal-identifier LinkedIn URLs (/in/ACwAA…), "API rejected my LinkedIn URLs", "convert LinkedIn URLs" |
Normalize person_linkedin_url before upload: drop encoded, recover vanity. |
Step 4 → "Normalize LinkedIn URLs" (this file) |
| "Make sure hashes are not double hashed" | Run the no-double-hash audit play before upload. | plays/audit-no-double-hash.play.ts |
| "enrich this list", "buy personal emails/hashes", "run the ladder" | Run the waterfall. Each layer only sees rows still missing a hash. | plays/enrich-audience-waterfall.play.ts |
| "Compare enriched versus unenriched" | Build both hash-only datasets and report lift. | plays/enrich-audience-waterfall.play.ts |
| "include phone numbers", "add phones" | Hash existing phones digits-only with country code. | shared/upload-failure-modes.md |
| "what was the match rate", "did it match" | Read contactIdInfo.matchRatePercentage, not the range enum. |
shared/upload-failure-modes.md |
| "put it in a sheet", "customer will upload" | Publish the validated file to Sheets; verify by row count. | shared/upload-failure-modes.md |
| "upload keeps failing", "422", "audience is locked" | Meta locks on write. Send the audience in one call. | shared/upload-failure-modes.md |
| "Upload to Google" | Validate hash-only rows, create Google audience, sync, readback. | plays/upload-google-hash-only-audience.play.ts |
| "Upload to Facebook and Google", "upload to FB/Google", "Meta + GAds" | Validate once, then upload to Google and Meta. | plays/upload-facebook-google-hash-only-audience.play.ts |
What ships with it
17 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.
- plays/audit-no-double-hash.play.ts 2.8 KB runs code
- plays/build-contactout-hash-pool.play.ts 9.7 KB runs code
- plays/build-hash-only-audience.play.ts 5.6 KB runs code
- plays/enrich-audience-waterfall.play.ts 16 KB runs code
- plays/evaluate-leadmagic-personal-email.play.ts 3.4 KB runs code
- plays/evaluate-personal-email-provider.play.ts 13 KB runs code
- plays/report-google-coverage-lift.play.ts 8.7 KB runs code
- plays/shared/audience-hash.ts 5.2 KB runs code
- plays/upload-facebook-google-hash-only-audience.play.ts 5.4 KB runs code
- plays/upload-google-hash-only-audience.play.ts 3.4 KB runs code
- recipes/enrich-and-upload-facebook-google.md 7.0 KB
- recipes/max-coverage-audience.md 11 KB
- recipes/sample-abm-segment-example.md 5.0 KB
- shared/audience-basics.md 4.5 KB
- shared/contactout-hash-pool.md 6.5 KB
- shared/upload-failure-modes.md 6.7 KB
- skill-metadata.json 3.3 KB
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 Changed acd9bcc883be
- 12d ago First seen · 498 lines · 133 tokens per session scan A 6e48aa149217
deepline-ads-audiences is a skill published in the GitHub repository getaero-io/gtm-eng-skills (58 stars, last pushed yesterday), licensed MIT. It adds 133 tokens to every session and 8,155 once invoked, about $0.0007 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 2026-08-30.
Other skills, from other repositories
inbound-lead-enrichment
Fills in missing data for inbound leads — researches the company, identifies the person's role and seniority, finds other stakeholders at the company, checks for existing CRM relationships, and updates the lead record. Produces enriched lead data ready for qualification or outreach. Tool-agnostic.
inbound-lead-qualification
Qualifies inbound leads against full ICP criteria — company size, industry, use case fit, role/seniority of the person. Checks CRM and existing customer base for duplicates and existing relationships. Outputs a scored CSV with qualification status, reasoning, and pipeline overlap flags. Tool-agnostic — works with any…
inbound-lead-triage
Triages all inbound leads from a given period — demo requests, free trial signups, content downloads, webinar registrations, chatbot conversations. Classifies by urgency, qualifies against ICP, enriches with context, and produces a prioritized action queue with recommended response for each lead. Tool-agnostic — works…
company-contact-finder
Find decision-makers at a specific company using Apollo, Crustdata, Fiber, and PDL people search via Gooseworks MCP. Given a company name and target titles, returns a list of contacts with name, title, LinkedIn URL, and location.
job-scraper
Search for job postings across LinkedIn and Indeed. Use when users want to find open roles, monitor hiring signals, identify companies hiring for specific positions, or research competitor hiring activity. Returns job title, company, location, salary, description, seniority level, and direct apply URLs. No login or…
apollo-lead-finder
Two-phase Apollo.io prospecting: free People Search to discover ICP-matching leads, then selective enrichment to reveal emails/phones (credits per contact). Creates Apollo lists. Deduplicates against existing contacts by LinkedIn URL.