# please tell me what skills are invoked by the 'deep-writing-mode' skill. don't just list them, te... ## Retrieval Notes - Session id: `019d0a59-28d0-7b11-998b-f43e1fefb4b7` - Source: `Agent Client fallback` - Last activity: `2026-03-20T08:25:41.538Z` ## My Notes <!-- Add your notes here. This section is preserved across syncs. --> ## Conversation ### User please tell me what skills are invoked by the 'deep-writing-mode' skill. don't just list them, tell me in detail what they are. ### Assistant I’m checking the `deep-writing-mode` skill definition and, because its files live outside the vault, I’m loading the `file-management` skill first so I can read them in the expected way. ### Assistant `deep-writing-mode` is a composite wrapper. I’m reading the skill files it pulls in so I can describe what each one actually does, not just repeat their names. ### Assistant I’ve got the mandatory core. I’m checking the four conditional add-ons as well, since `deep-writing-mode` treats them as part of the bundle when the session calls for them. ### Assistant `deep-writing-mode` is a composite launcher, not a writing method in its own right. Its job is to preload a stack of other skills so that manuscript work is governed by a consistent voice, formatting discipline, epistemic discipline, and source-handling protocol. The definition is in [SKILL.md](/Users/nickyoung/.codex/skills/deep-writing-mode/SKILL.md). It always invokes these four, in this order: 1. `nick-analytic-voice` This is the core voice-governing skill. It tells the model to write in Nick’s academic philosophical prose style: British English, no contractions, longer argumentative sentences doing the thinking, shorter ones landing the point, first-person ownership without bluster, and direct engagement with interlocutors rather than vague literature-summary. It is also strongly anti-LLM in a specific way: it forbids announcement phrases, empty evaluatives, generic “this establishes…” meta-commentary, flat chains of short sentences, decorative metaphors, and overuse of stock moves. It requires grounding in reference passages before drafting and a post-draft checklist afterwards. In practice, this is the skill that decides what the prose should sound like and what kinds of stylistic failure count as unacceptable. See [SKILL.md](/Users/nickyoung/.codex/skills/nick-analytic-voice/SKILL.md). 2. `writing-standards` This is the formatting and convention layer. It handles quotation-mark usage, italics, bold restraint, punctuation with quotations in UK style, note-title conventions, and when formal versus informal style is appropriate. Its main function is not “voice” in the richer sense, but mechanical consistency: double quotes for direct quotation, single quotes for scare quotes or word-mention, italics for titles and technical terms on first introduction, almost no bold, UK logical punctuation, and APA 7th for citations. In practice, this skill stops prose from drifting into inconsistent stylistic habits even if the argument itself is good. See [SKILL.md](/Users/nickyoung/.codex/skills/writing-standards/SKILL.md). 3. `epistemic-discipline` This skill governs how Nick’s developing ideas are represented. Its central concern is not prose quality but intellectual hygiene. It forbids the model from imposing hierarchy on in-progress thinking by calling something “the main point”, “the central claim”, “the key insight”, and so on unless Nick himself has explicitly ranked it. It insists on flat presentation of parallel threads, explicit marking of when organisation or interpretation comes from the assistant rather than Nick, and careful attribution when any priority or focus is claimed. In practice, this skill prevents the assistant from prematurely collapsing an exploratory research space into a false structure. It is especially about preserving openness rather than manufacturing clarity too early. See [SKILL.md](/Users/nickyoung/.codex/skills/epistemic-discipline/SKILL.md). 4. `source-work` This is the scholarly-integrity layer. It imposes an extraction-first workflow for academic sources: do not quote from memory, extract the text first, read the extraction, work only from that extraction, verify every quotation and attribution, and distinguish clearly between what the source says, what is interpretation, and what is speculation. It also includes format-specific rules for PDFs, EPUBs, and text files, plus a warning against naive reading of large PDFs. In practice, this skill exists to stop plausible fabrication: if a source is being discussed, the model is supposed to have the actual text in front of it, not rely on recollection or synthesis alone. See [SKILL.md](/Users/nickyoung/.codex/skills/source-work/SKILL.md). `deep-writing-mode` can also conditionally invoke four more skills when the session calls for them: - `depth-audit` This is for checking whether prose is genuinely doing philosophical work or merely describing what it claims to be doing. It has a rich diagnostic framework for shallow writing: “described but not made”, “asserted without earning”, “named but not developed”, “summary substituting for argument”, “list substituting for development”, and “quotation without analysis”. Its standard of comparison is Nick’s published papers, and it insists on paragraph-by-paragraph comparison against those. This is not about style or source accuracy; it is about whether the prose actually thinks on the page. See [SKILL.md](/Users/nickyoung/.codex/skills/depth-audit/SKILL.md). - `nick-philosophical-prose` This is closely related to `nick-analytic-voice`, but more specifically aimed at turning a prior philosophical conversation into finished prose without losing the fine-grained distinctions worked out in discussion. Its emphasis is on preserving specificity: not saying “the argument turns on a distinction”, but actually stating the distinction; not gesturing at an example, but fully working it through. If `nick-analytic-voice` is the general voice profile, this one is the conversation-to-draft transfer protocol. See [SKILL.md](/Users/nickyoung/.codex/skills/nick-philosophical-prose/SKILL.md). - `twork` This is for working through `%%comments%%` inside drafts. It is exploratory rather than implementation-focused: for each flagged issue it is supposed to analyse what the comment is really worried about, generate multiple options, explain tradeoffs, and stay voice-aware. It is meant for cases where the text contains unresolved local problems and Nick wants deep engagement with those problems before deciding how to revise. See [SKILL.md](/Users/nickyoung/.codex/skills/twork/SKILL.md). - `contemplate` This is the heavy deliberation mode. It is not specifically a writing skill, but a reasoning-style override for sessions where deep exploratory thinking is needed before or around the writing. It requires extended visible deliberation, multiple hypotheses, backtracking, and a very slow movement toward any conclusion. In the `deep-writing-mode` bundle it functions as the “do the thinking first” escalator when the intellectual issue itself is not yet settled. See [SKILL.md](/Users/nickyoung/.codex/skills/contemplate/SKILL.md). So the short version is: `deep-writing-mode` always loads voice, formatting, epistemics, and source-discipline first; then it adds depth-audit, conversation-to-prose rendering, comment-work, or deep deliberation only if the session specifically needs those.