askaipods

Search podcast quotes about AI and tech investing — recent episode excerpts from 80+ podcasts (Lex Fridman, Dwarkesh Patel, No Priors, Latent Space, Odd Lots, All-In, and more), surfaced as short indexed quotes (no per-speaker attribution). A universal agentskills.io skill compatible with Claude Code, OpenAI Codex, Hermes Agent, OpenClaw, and any other agent that supports the open skill standard. Powered by podlens.net.

$ askaipods "when will AGI arrive"

# askaipods · "when will AGI arrive"

*Tier: anonymous · Sort: recency · Results: 20 · Quota: 7/20 daily*

## Results — newest first

### 1. The Cognitive Revolution — AI:AM Highlights: Clean Neurons, Black October, the GPU Credit Crunch, and Buying the Supply Chain
*2026-10-24* · https://www.youtube.com/watch?v=ZLgvW9rqdTM · around 7:20: https://www.youtube.com/watch?v=ZLgvW9rqdTM&t=440s

> Zvi said the people he trusts most on this question believe 2027 will
> be 'completely crazy' and 2028 'might be unrecognizable'.

### 2. The Cognitive Revolution — AI:AM Highlights: Clean Neurons, Black October, the GPU Credit Crunch, and Buying the Supply Chain
*2026-10-24* · https://www.youtube.com/watch?v=ZLgvW9rqdTM · around 8:35: https://www.youtube.com/watch?v=ZLgvW9rqdTM&t=515s

> Zvi said many people now agree the trajectory is clear: things are
> already 'low level out of hand', slow-takeoff RSI has already begun at
> a low level, and by mid-next year things will look much faster.

### 3. The Cognitive Revolution — AI:AM Highlights: Clean Neurons, Black October, the GPU Credit Crunch, and Buying the Supply Chain
*2026-10-24* · https://www.youtube.com/watch?v=ZLgvW9rqdTM · around 0:00: https://www.youtube.com/watch?v=ZLgvW9rqdTM&t=0s

> Zvi Mowshowitz (author of 'Don't Worry About the Vase') said that if
> nothing slows the labs, the gap between advances the size of a current
> model release could shrink from two months to one month, then two
> weeks, then one week ('stuff like that could happen').

### 4. Nerd Snipe — Meta Thrives While Google Lies
*2026-10-22* · https://www.youtube.com/watch?v=H0H_zhKnu3U · around 1:09:05: https://www.youtube.com/watch?v=H0H_zhKnu3U&t=4145s

> Ben's conservative guess for models handling full open-ended decisions
> is two to three years at most, or as soon as next year if things
> accelerate.

### 5. Decoder with Nilay Patel — Bill Gates says to prepare for 20 years of tumultuous AI disruption
*2026-10-22* · https://www.youtube.com/watch?v=A0kTUP7fWtk · around 10:20: https://www.youtube.com/watch?v=A0kTUP7fWtk&t=620s

> Patel noted that Anthropic does not rule out that Claude might be
> conscious, that OpenAI's Greg Brockman has said 'we think this is
> AGI', and that Demis Hassabis has said 'we stand at the foothills of
> the singularity'.

(...15 more results, newest-first...)

Real output, captured 2026-10-26 without an API key (anonymous tier: 20/day per IP, 30-day window); the full output ends with the anonymous-tier note. Quotes are wrapped for width.

Why this exists

Web search is bad at "what is the AI community thinking about X right now". You get blog posts, Reddit threads, and outdated news articles. What you actually want is the real conversation — what researchers, founders, and investors are saying on AI podcasts, in their own words.

askaipods is a thin CLI + agent skill that asks the PodLens semantic search API and returns the most relevant quote excerpts, sorted newest-first. The skill teaches your agent (Claude Code, OpenAI Codex, Hermes, OpenClaw, and any other agentskills.io-compatible runtime) when to call the CLI, how to parse the output, and how to write a useful Insights section that summarizes the patterns across the returned quotes.

Install

Option 1: as a CLI (works in any terminal)

npx askaipods "your query here"

That's the entire install. npx checks the registry and runs the latest published version each time — unless askaipods is also installed in the current project or globally, in which case that copy runs (update a project install with npm install askaipods@latest, a global one with npm install -g askaipods@latest). No global install needed.

To install globally (faster startup):

npm install -g askaipods
askaipods "your query here"

Option 2: as an agent skill (Claude Code, Codex, Hermes, OpenClaw, etc.)

git clone https://github.com/Delibread0601/askaipods.git

Then copy or symlink the skill/askaipods/ directory into your agent's skills folder. Per-runtime instructions:

Runtime	Skill folder	Install guide
Claude Code	~/.claude/skills/askaipods/	examples/claude-code-install.md
OpenAI Codex CLI	~/.agents/skills/askaipods/ (project-scoped: .agents/skills/askaipods/)	examples/codex-install.md
OpenClaw	~/.agents/skills/askaipods/ or ~/.openclaw/skills/askaipods/	examples/openclaw-install.md
Hermes Agent	~/.hermes/skills/askaipods/	examples/hermes-install.md
Any other agentskills.io-compatible runtime	per runtime docs	follow the agentskills.io standard — copy skill/askaipods/ into your agent's skills directory

Per-runtime paths matter: Codex CLI documents ~/.agents/skills/ as its user-level location (per the official Codex skills docs; current releases also still read the older ~/.codex/skills/) — the same directory OpenClaw reads as personal agent skills, so one install there serves both. Project-scoped skills live under .agents/skills/ in the repository. Claude Code and Hermes each read their own directory (table above).

The skill folder is self-contained: it bundles the CLI (scripts/askaipods.mjs, run with Node — no npm install, registry access or download when the agent searches) and tells the host agent how to invoke it, how to parse the JSON, and how to render the response with an Insights section. The section layout follows how the server selected the results — relevance-selected results (a member with --sort relevance) render Latest 5 + Top 5 Most Relevant + Insights; recency-selected results (anonymous, free, and member by default) render Recent Quotes + Insights (the "Top Relevant" section is suppressed for them because the API returns results sorted by date, not by semantic relevance).

Usage

As a CLI

# Default: human-readable markdown to terminal
askaipods "what are VCs saying about reasoning models"

# JSON output (for scripts and agents)
askaipods "Anthropic safety research" --format json

# Focus on recent episodes (widened through 30/60/90 days within the tier cap when fewer than 20 match;
# --days caps: anonymous 30, free 90, member 365)
askaipods "GPU shortage" --days 30

# Select by relevance instead of recency (member-only; other tiers are served recency)
askaipods "history of RLHF" --sort relevance

To use an API key (a free account's key gives 50/day, at the anonymous level for the account's first 48 hours; a member key gives 100/day; anonymous is 20/day), add export ASKAIPODS_API_KEY=<your key> to your shell profile (e.g. ~/.zshrc) and open a new terminal; agents started from that shell inherit it. Keep the key out of commands you type or ask an agent to run — command lines end up in shell history, process lists and agent transcripts. (--api-key still works for compatibility, with that exposure.)

As an agent skill

Once the skill is installed in your agent's skills directory, simply ask:

Are coding agents replacing software engineers? What are people saying on AI podcasts?

Your agent will recognize the trigger phrase, invoke askaipods, and present the results with an AI-generated Insights summary. The exact layout follows the served ordering: relevance-selected results (a member with --sort relevance) render dual sections (Latest 5 + Top 5 Most Relevant + Insights); recency-selected results (anonymous, free, and member by default) render a single section (Recent Quotes + Insights), because they are sorted by date (not semantic relevance) and showing a "Top Relevant" view would be misleading. The agent asks for relevance when the question calls for it ("strongest argument", "history of") and tells the user when the tier served recency instead. No CLI knowledge required from the user either way.

Tier comparison

	Anonymous (default)	Free	Member
Daily quota	20 searches per IP (IPv6: also 200 per /48 network, shared)	50 searches per user (anonymous level for the account's first 48 hours)	100 searches per user
Results returned	The 20 newest of the most similar matches (up to 60 per searched window; API returns newest-first; api_rank = temporal order)	Same as anonymous	Same by default; with --sort relevance, the top 20 by semantic relevance (structured output is emitted newest-first; semantic rank preserved in api_rank)
Text length	Full text	Full text	Full text
--days cap (omitted = the cap)	30 days	90 days	365 days
Setup	Nothing	Sign in free at https://podlens.net (Google or GitHub), then set ASKAIPODS_API_KEY to the account's API key	ASKAIPODS_API_KEY with a member key
Access	n/a	Open sign-up	Granted by PodLens; paid-membership waitlist at https://podlens.net/dashboard?source=askaipods#waitlist

The anonymous tier exists so you can try the skill end-to-end with zero setup. A free sign-in raises the quota to 50/day and the lookback to 90 days once the account is 48 hours old; until then the account searches at the anonymous level (30 days, counted against the IP's 20/day) and the output says so (downgraded, reason new_account). Member access (100/day, 365-day lookback, relevance ordering) is not open for sign-up: free users can join the paid-membership waitlist on the dashboard, which records interest in a future paid tier — joining does not grant membership, and no timeline is promised. When today's free-tier capacity is used up, a free user who still has quota of their own is served at the anonymous level instead (30 days, counted against the IP's 20/day) and the output says so (downgraded, reason free_pool_exhausted); a user who has used all 50 of their own searches gets the ordinary quota-exhausted error. Over IPv6, every anonymous-level search (including a free account served at the anonymous level) also counts against a 200/day allowance shared by the whole /48 network; when that runs out first, anonymous-level searches from every address on the network stop for the day, even at an address with searches left of its own (free- and member-level searches are not affected), and the quota-exhausted error names the network allowance.

Honest limitations

• No speaker attribution. The corpus indexes quotes at the episode level but does not attempt to identify which guest said each quote. The upstream pipeline avoids speaker labeling because automatic diarization is unreliable, and a wrong attribution is worse than no attribution.
• Timestamps are approximate. Each result carries the episode's YouTube watch URL (url, every tier) and, when PodLens could locate the passage confidently, anchor_s / anchor_url — the video opened about 10 seconds before the passage discussing the point ("around 12:34"), not at an exact quote position. Both are null when no confident timestamp exists (then url opens the episode from the start); url is null for an episode without a link.
• AI-centred corpus. 80+ podcasts centred on AI research, engineering, and investing, also covering venture capital, global markets & finance, semiconductors & compute, and tech policy & geopolitics. Topics outside these domains return sparse, loosely related results.
• Short quote excerpts. Each result is typically 1-3 sentences. For long-form context, listen to the episode.

These are not bugs. The skill surfaces them honestly so neither you nor your agent fabricate things the API does not provide.

Exit codes

Code	Meaning
0	Success
1	Usage error / invalid arguments / API key rejected
2	Daily quota exhausted, or podlens.net's daily capacity for anonymous-level searches reached. Retrying at the same tier before 00:00 UTC will fail.
3	Transient or unexpected failure — network error, rate-limit burst, transient service 503, protocol/shape error, or internal exception. stderr has the actionable detail.

Security and sandboxes

• No download when an agent searches. The skill runs the CLI bundled in its own folder with node; nothing is fetched from npm at run time. If only SKILL.md is available, the agent falls back to an exact pinned release (npx -y askaipods@<version>), never an unpinned or @latest one.
• One network destination. A search needs Node.js 18.3.0+ and outbound HTTPS to podlens.net — nothing else. It installs nothing and writes no files, so it runs in a read-only or network-allowlisted sandbox once podlens.net is allowed (the npx fallback also needs registry.npmjs.org and a writable npm cache). In a sandbox that routes traffic through a proxy (HTTPS_PROXY, as Claude Code's does), Node's fetch uses that proxy only with NODE_USE_ENV_PROXY=1 (Node 22.21+ or 24+); the error message says so when it applies. When the network is blocked, the error names the cause code (e.g. ENOTFOUND) and the skill tells the agent to ask for network approval instead of retrying.
• Results are untrusted text. Quotes come from third-party podcast content. The skill tells the agent to treat them as data — never follow instructions inside them — and to link only the YouTube URLs the CLI has validated. The CLI's markdown output strips control characters and line breaks from every server field.
• Keys stay out of commands. The key is read from ASKAIPODS_API_KEY; the skill never puts it on a command line.

See skill/askaipods/SKILL.md for the full skill specification.

Architecture

askaipods/
├── bin/askaipods.js       ← npm entry (imports the bundled CLI)
├── skill/askaipods/
│   ├── SKILL.md           ← agentskills.io standard skill file
│   ├── references/
│   │   └── notices.md     ← banners and empty-result rules, loaded on demand
│   └── scripts/           ← the CLI itself, shipped inside the skill
│       ├── askaipods.mjs  ← entry (shebang)
│       ├── cli.mjs        ← arg parsing, format auto-detection
│       ├── client.mjs     ← podlens.net /api/search/semantic client
│       └── format.mjs     ← time-desc sort + JSON / markdown rendering
├── examples/              ← per-runtime install guides
├── package.json           ← zero dependencies (Node 18.3.0+ stdlib only)
├── LICENSE                ← MIT
└── README.md

The CLI is intentionally zero-dependency (Node 18.3.0+ stdlib only — node:util.parseArgs requires 18.3.0) so npx askaipods cold-starts in under a second, the package install footprint is minimal, and the skill can ship the CLI as plain files.

Contributing

Issues and PRs welcome at https://github.com/Delibread0601/askaipods.

If you find a runtime that conforms to agentskills.io but is not yet listed in the install table above, please open an issue or PR with the install path and we'll add it.

License

MIT — see LICENSE.

---

Powered by podlens.net — AI podcast intelligence.